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。

What structural task representation enables multi-stage, in-the-wild, bimanual, reactive manipulation? Introducing ReKep: LVM to label keypoints & VLM to write keypoint-based constraints, solve w/ optimization for diverse tasks, w/o task-specific training or env models. 🧵👇

Wenlong Huang

5,697 subscribers

190,914 次观看 • 1 年前 •via X (Twitter)

科学技术教育

Anya Rossi• Live Now

Private livecam show

11 条评论

Wenlong Huang 的头像

Wenlong Huang1 年前

Relational Keypoint Constraints (ReKep) represent tasks as seq of keypoint relations. Eg in pouring task: pull together gripper kp & handle kp -> keep handle & spout kp at same height (avoid spillage) -> align spout & cup kp -> handle & spout kp form a tilting angle to pour. 2/N

Wenlong Huang 的头像

Wenlong Huang1 年前

ReKeps are Python functions mapping kp to costs w/ NumPy operations, specifying relations b/w robot, obj, and obj parts. While each kp has only (x,y,z), multiple kp can specify SO(3) rotations, vectors, surfaces, volumes to capture rich geometric structures in manipulation. 3/N

Wenlong Huang 的头像

Wenlong Huang1 年前

By sequencing multiple ReKeps, our framework organically integrates high-level task planning with dense low-level actions as a unified continuous optimization problem. With tracked keypoints, this enables rapid backtracking & replanning behaviors both within/across stages. 4/N

Wenlong Huang 的头像

Wenlong Huang1 年前

ReKep can also be fully automated w/ foundation models for in-the-wild task execution. We use large vision models (SAM+DINOv2) to identify keypoints, overlay on image, and prompt VLM (GPT-4o) to write a seq of ReKep constraints based on the task instruction. 5/N

Wenlong Huang 的头像

Wenlong Huang1 年前

We test ReKep on two setups – in-the-wild and bimanual. It can perform diverse 6-12 DoF tasks w/ a perception-action loop at 10 Hz. It can also fold diff. clothes with diff. (human-like) strategies. And entire pipeline does not require task-specific training or env models. 6/N

Wenlong Huang 的头像

Wenlong Huang1 年前

ReKep is implemented upon common packages like SciPy, and the code has also been open-sourced! The code runs on BEHAVIOR ( a large-scale benchmark with diverse scenes and objects, so you can easily try ReKep without setting up a real robot. 7/N

Wenlong Huang 的头像

Wenlong Huang1 年前

Project website: Walkthrough video: Paper: Code: Work done together with @chenwang_j, @YunzhuLiYZ, @RuohanZhang76, and @drfeifei. (N/N)

Huazhe Harry Xu 的头像

Huazhe Harry Xu1 年前

Congrats, @wenlong_huang ! This is impressive!

Wenlong Huang 的头像

Wenlong Huang1 年前

Thanks Huazhe!!

Karol Hausman 的头像

Karol Hausman1 年前

Cool work @wenlong_huang 👏

Wenlong Huang 的头像

Wenlong Huang1 年前

Thanks Karol!!

相关视频

How to harness foundation models for *generalization in the wild* in robot manipulation? Introducing VoxPoser: use LLM+VLM to label affordances and constraints directly in 3D perceptual space for zero-shot robot manipulation in the real world! 🌐 🧵👇

How to harness foundation models for generalization in the wild in robot manipulation? Introducing VoxPoser: use LLM+VLM to label affordances and constraints directly in 3D perceptual space for zero-shot robot manipulation in the real world! 🌐 🧵👇

Wenlong Huang

293,910 次观看 • 3 年前

Learned visuomotor policies are notoriously fragile, they break with changes in conditions like lighting, clutter, or object variations amongst other things. In Yunchu @ CoRL2025's latest work, we asked whether we could get these policies to be robust and generalizable with a clever choice of visual representation! The argument we made was - we want a choice of visual representation that specifically adapts to be sufficient, yet minimal for the task at hand. We thought about it from the perspective of flexible, key-point based representations. The key question becomes - how do we choose a sufficient, task-specific, yet minimal set of keypoints as a representation for policy learning. Yunchu proposes a neat way of automatically selecting task-relevant keypoints using a standard supervised learning objective, and using this for robust policy learning. This is largely under the same assumptions as behavior cloning, but with huge gains on robustness. Let’s understand how, 🧵 (1/8)

Learned visuomotor policies are notoriously fragile, they break with changes in conditions like lighting, clutter, or object variations amongst other things. In Yunchu @ CoRL2025's latest work, we asked whether we could get these policies to be robust and generalizable with a clever choice of visual representation! The argument we made was - we want a choice of visual representation that specifically adapts to be sufficient, yet minimal for the task at hand. We thought about it from the perspective of flexible, key-point based representations. The key question becomes - how do we choose a sufficient, task-specific, yet minimal set of keypoints as a representation for policy learning. Yunchu proposes a neat way of automatically selecting task-relevant keypoints using a standard supervised learning objective, and using this for robust policy learning. This is largely under the same assumptions as behavior cloning, but with huge gains on robustness. Let’s understand how, 🧵 (1/8)

Abhishek Gupta

11,355 次观看 • 1 年前

Introducing Skills in SciSpace Agent. Specialized modes built for specific research workflows. Describe the task, and the right workflow activates automatically. Yes, the right Skill gets loaded based on what you need. Built to simplify multi-step research work and reduce friction around execution. Skills are now live. Try it inside SciSpace Agent.

Introducing Skills in SciSpace Agent. Specialized modes built for specific research workflows. Describe the task, and the right workflow activates automatically. Yes, the right Skill gets loaded based on what you need. Built to simplify multi-step research work and reduce friction around execution. Skills are now live. Try it inside SciSpace Agent.

SciSpace

43,422 次观看 • 4 个月前

Is VideoGen starting to become good enough for robotic manipulation? 🤖 Check out our recent work, RIGVid — Robots Imitating Generated Videos — where we use AI-generated videos as intermediate representations and 6-DoF motion retargeting to guide robots in diverse manipulation tasks: pouring, wiping, mixing, and more. 🔗 Key takeaways: - VideoGen starts to become good enough for robotics - As the field progresses, we are expecting much better results in the coming years - Depending on whether video prediction models take actions or not (VideoGen vs Action-Conditioned Video Prediction), there are different ways to use them. - Controllability & steerability are still issues In the paper, we explore: – How do different video generation models compare for robotic imitation? – Can generated videos replace real videos for imitation? – What causes failure of imitation given high-quality videos? – How does imitating from video compare with other representations (e.g., keypoint constraints like ReKep)? 🎥 Watch the video for (1) AI-generated inputs, (2) robot executions, and (3) the 3D intermediate representation bridging the embodiment gap.

Is VideoGen starting to become good enough for robotic manipulation? 🤖 Check out our recent work, RIGVid — Robots Imitating Generated Videos — where we use AI-generated videos as intermediate representations and 6-DoF motion retargeting to guide robots in diverse manipulation tasks: pouring, wiping, mixing, and more. 🔗 Key takeaways: - VideoGen starts to become good enough for robotics - As the field progresses, we are expecting much better results in the coming years - Depending on whether video prediction models take actions or not (VideoGen vs Action-Conditioned Video Prediction), there are different ways to use them. - Controllability & steerability are still issues In the paper, we explore: – How do different video generation models compare for robotic imitation? – Can generated videos replace real videos for imitation? – What causes failure of imitation given high-quality videos? – How does imitating from video compare with other representations (e.g., keypoint constraints like ReKep)? 🎥 Watch the video for (1) AI-generated inputs, (2) robot executions, and (3) the 3D intermediate representation bridging the embodiment gap.

Yunzhu Li

16,540 次观看 • 1 年前

What if robots could improve themselves by learning from their own failures in the real-world? Introducing 𝗣𝗟𝗗 (𝗣𝗿𝗼𝗯𝗲, 𝗟𝗲𝗮𝗿𝗻, 𝗗𝗶𝘀𝘁𝗶𝗹𝗹) — a recipe that enables Vision-Language-Action (VLA) models to self-improve for high-precision manipulation tasks. PLD couples real-world residual reinforcement learning with standard supervised fine-tuning — letting robots discover, recover, and distill their own data flywheel. Quick 🧵

What if robots could improve themselves by learning from their own failures in the real-world? Introducing 𝗣𝗟𝗗 (𝗣𝗿𝗼𝗯𝗲, 𝗟𝗲𝗮𝗿𝗻, 𝗗𝗶𝘀𝘁𝗶𝗹𝗹) — a recipe that enables Vision-Language-Action (VLA) models to self-improve for high-precision manipulation tasks. PLD couples real-world residual reinforcement learning with standard supervised fine-tuning — letting robots discover, recover, and distill their own data flywheel. Quick 🧵

Wenli Xiao

185,017 次观看 • 8 个月前

Today, we publicly released RoboCasa365, a large-scale simulation benchmark for training and systematically evaluating generalist robot models. Built upon our original RoboCasa framework, it offers: • 2,500 realistic kitchen environments; • 365 everyday tasks (basic skills + long-horizon mobile manipulation); • Over 3,200 objects with many articulated fixtures/appliances. All are designed for fully controlled, reproducible benchmarking of robotic policies. Progress in robotic foundation models is real. But it’s still hard to answer basic questions like: How close are we to general-purpose autonomy? What factors drive generalization? What are the model/data scaling curves like? Real-world eval is slow and noisy, and existing sims (like LIBERO, which we built 3 years ago) often lack sufficient task and scene diversity. This benchmark comes with 2,200+ hours of demonstrations and 500K+ trajectories to support studies of multi-task training, pretraining, and continual learning at scale. Check it out at

Today, we publicly released RoboCasa365, a large-scale simulation benchmark for training and systematically evaluating generalist robot models. Built upon our original RoboCasa framework, it offers: • 2,500 realistic kitchen environments; • 365 everyday tasks (basic skills + long-horizon mobile manipulation); • Over 3,200 objects with many articulated fixtures/appliances. All are designed for fully controlled, reproducible benchmarking of robotic policies. Progress in robotic foundation models is real. But it’s still hard to answer basic questions like: How close are we to general-purpose autonomy? What factors drive generalization? What are the model/data scaling curves like? Real-world eval is slow and noisy, and existing sims (like LIBERO, which we built 3 years ago) often lack sufficient task and scene diversity. This benchmark comes with 2,200+ hours of demonstrations and 500K+ trajectories to support studies of multi-task training, pretraining, and continual learning at scale. Check it out at

Yuke Zhu

23,977 次观看 • 4 个月前

chatGPT custom instructions is INSANE. You can build full agents!! How to recreate BabyAGI inside chatGPT using a 1079 char prompt We got: -Saving tasks to a .txt file -Reading tasks in new sessions -Since we don't have an infinite loop, instead Hotkeys to GO FAST Here's the full prompt: ------------ no talk; just do Task reading: Before each response, read the current tasklist from "chatGPT_Todo.txt". Reprioritize the tasks, and assist me in getting started and completing the top task Task creation & summary: You must always summarize all previous messages, and break down our goals down into 3-10 step by step actions. Write code and save them to a text file named "chatGPT_Todo.txt". Always provide a download link. Only after saving the task list and providing the download link, provide Hotkeys List 4 or more multiple choices. Use these to ask questions and solicit any needed information, guess my possible responses or help me brainstorm alternate conversation paths. Get creative and suggest things I might not have thought of prior. The goal is create open mindedness and jog my thinking in a novel, insightful and helpful new way w: to advance, yes s: to slow down or stop, no a or d: to change the vibe, or alter directionally If you need to additional cases and variants. Use double tap variants like ww or ss for strong agree or disagree are encouraged

chatGPT custom instructions is INSANE. You can build full agents!! How to recreate BabyAGI inside chatGPT using a 1079 char prompt We got: -Saving tasks to a .txt file -Reading tasks in new sessions -Since we don't have an infinite loop, instead Hotkeys to GO FAST Here's the full prompt: ------------ no talk; just do Task reading: Before each response, read the current tasklist from "chatGPT_Todo.txt". Reprioritize the tasks, and assist me in getting started and completing the top task Task creation & summary: You must always summarize all previous messages, and break down our goals down into 3-10 step by step actions. Write code and save them to a text file named "chatGPT_Todo.txt". Always provide a download link. Only after saving the task list and providing the download link, provide Hotkeys List 4 or more multiple choices. Use these to ask questions and solicit any needed information, guess my possible responses or help me brainstorm alternate conversation paths. Get creative and suggest things I might not have thought of prior. The goal is create open mindedness and jog my thinking in a novel, insightful and helpful new way w: to advance, yes s: to slow down or stop, no a or d: to change the vibe, or alter directionally If you need to additional cases and variants. Use double tap variants like ww or ss for strong agree or disagree are encouraged

Nick Dobos

1,435,875 次观看 • 3 年前

Most motion papers tailor one controller to one specific task. This year at SIGGRAPH, our research team asks: can motor control itself be pretrained and reused? Generative Pretrained Controllers, or GPC, turn motor skills into a vocabulary of discrete tokens and train a transformer-based generative controller through next-token prediction. Just like GPT, the same pretrained controller can then be fine-tuned to solve new tasks. Trained on 600+ hours of motion, GPC runs in real-time inside a physics simulation, producing natural and physically grounded behaviors for interactive control.

Most motion papers tailor one controller to one specific task. This year at SIGGRAPH, our research team asks: can motor control itself be pretrained and reused? Generative Pretrained Controllers, or GPC, turn motor skills into a vocabulary of discrete tokens and train a transformer-based generative controller through next-token prediction. Just like GPT, the same pretrained controller can then be fine-tuned to solve new tasks. Trained on 600+ hours of motion, GPC runs in real-time inside a physics simulation, producing natural and physically grounded behaviors for interactive control.

NVIDIA AI

175,153 次观看 • 27 天前

Introducing EdgeBench, a benchmark designed to study how agents learn from environments over at least 12~72-hour runs. We find that performance follows a log-sigmoid function of environment interaction time with high precision. EdgeBench is built with three ingredients: - 🌍 Real & Diverse: 134 real-world tasks across 6 task categories, spanning scientific problems, professional knowledge work, software engineering, optimization, formal math, and games. - ⏳ Ultra-Long-Horizon: Each task supports 12–72 hours of agent work. Recorded human effort averages 57.2 hours. - 🔁 Informative Feedback: Agents receive real-world feedback for continuous improvement. After 38,000 hours of agent runs on EdgeBench, a scaling law for learning from environments emerges: - 📈 As agents interact with task environments over time, their aggregate performance is precisely fit by a log-sigmoid function. - 🧠 This phenomenon can be explained by an elegant theory of graph exploration. We are releasing an initial 51 of the 134 tasks, together with the full evaluation framework, to help advance long-horizon agent research. Check our blog & paper for more findings! Blog Paper GitHub Dataset Details below 👇🧵

Introducing EdgeBench, a benchmark designed to study how agents learn from environments over at least 12~72-hour runs. We find that performance follows a log-sigmoid function of environment interaction time with high precision. EdgeBench is built with three ingredients: - 🌍 Real & Diverse: 134 real-world tasks across 6 task categories, spanning scientific problems, professional knowledge work, software engineering, optimization, formal math, and games. - ⏳ Ultra-Long-Horizon: Each task supports 12–72 hours of agent work. Recorded human effort averages 57.2 hours. - 🔁 Informative Feedback: Agents receive real-world feedback for continuous improvement. After 38,000 hours of agent runs on EdgeBench, a scaling law for learning from environments emerges: - 📈 As agents interact with task environments over time, their aggregate performance is precisely fit by a log-sigmoid function. - 🧠 This phenomenon can be explained by an elegant theory of graph exploration. We are releasing an initial 51 of the 134 tasks, together with the full evaluation framework, to help advance long-horizon agent research. Check our blog & paper for more findings! Blog Paper GitHub Dataset Details below 👇🧵

Deyao Zhu

356,702 次观看 • 25 天前

We benchmarked leading multimodal foundation models (GPT-4o, Claude 3.5 Sonnet, Gemini, Llama, etc.) on standard computer vision tasks—from segmentation to surface normal estimation—using standard datasets like COCO and ImageNet. These models have made remarkable progress; however, it is unclear exactly where they stand in terms of understanding vision in detail. Especially when it comes to tasks beyond question-answering. How well do they understand an object's segments or geometry? Our analyses yield an assessment that is quantitatively and qualitatively detailed and is compatible with evaluations developed in the field of computer vision over the past decades. Observed trends: 🔹 The foundation models consistently underperform task-specific SOTA models across all tasks. However, they are respectable generalists, which is remarkable as they are presumably trained primarily on image-text-based tasks. 🔹 They perform semantic tasks notably better than geometric ones. 🔹 GPT-4o performs the best among non-reasoning models, getting the top position in 4 out of 6 tasks. 🔹 Reasoning models, e.g., o3, show improvements in geometric tasks. 🔹 The 'image generation' models, e.g., GPT-40 Image Generation, which have been natively trained multimodally, exhibit quirks. E.g., hallucinated objects, misalignment between the input and output, etc. 🔹 While the prompting techniques affect performance, better models exhibit less sensitivity to variations in prompts. We control for the variance introduced by the prompting methods in our experiments. 🌐 Detailed analyses, visualizations: ⌨️ code: 🧵 1/n

We benchmarked leading multimodal foundation models (GPT-4o, Claude 3.5 Sonnet, Gemini, Llama, etc.) on standard computer vision tasks—from segmentation to surface normal estimation—using standard datasets like COCO and ImageNet. These models have made remarkable progress; however, it is unclear exactly where they stand in terms of understanding vision in detail. Especially when it comes to tasks beyond question-answering. How well do they understand an object's segments or geometry? Our analyses yield an assessment that is quantitatively and qualitatively detailed and is compatible with evaluations developed in the field of computer vision over the past decades. Observed trends: 🔹 The foundation models consistently underperform task-specific SOTA models across all tasks. However, they are respectable generalists, which is remarkable as they are presumably trained primarily on image-text-based tasks. 🔹 They perform semantic tasks notably better than geometric ones. 🔹 GPT-4o performs the best among non-reasoning models, getting the top position in 4 out of 6 tasks. 🔹 Reasoning models, e.g., o3, show improvements in geometric tasks. 🔹 The 'image generation' models, e.g., GPT-40 Image Generation, which have been natively trained multimodally, exhibit quirks. E.g., hallucinated objects, misalignment between the input and output, etc. 🔹 While the prompting techniques affect performance, better models exhibit less sensitivity to variations in prompts. We control for the variance introduced by the prompting methods in our experiments. 🌐 Detailed analyses, visualizations: ⌨️ code: 🧵 1/n

Amir Zamir

73,074 次观看 • 1 年前

Israel-based Mentee Robotics has demonstrated a logistics workflow: two MenteeBot V3 humanoids work autonomously to pick and place totes. A Modular Agent System is preferred because it favors real-world robustness and lower compute needs over the End-to-End VLA model. Its architecture is composed of three components: - LLM Planner: Converts instructions into executable Robotic API Language code for reliable task decomposition and error handling. - Perception Stack: Uses pre-trained models (NeRF/3DGS, distilled vision) for scene understanding and navigation. - Control Policies: Reinforcement Learning (RL) models, trained at scale via Sim2Real, generate motor commands, enabling high-accuracy mobile manipulation. Crucially, the robot learns new tasks from a single demonstration in hours. Object tracking uses 3D geometry (STL/URDF) tracked in the video to define the RL reward function. Training is optimized using 'Automatic Curriculum Learning', which autonomously adjusts task difficulty based on robot performance, eliminating manual engineering. All computation runs onboard.

Israel-based Mentee Robotics has demonstrated a logistics workflow: two MenteeBot V3 humanoids work autonomously to pick and place totes. A Modular Agent System is preferred because it favors real-world robustness and lower compute needs over the End-to-End VLA model. Its architecture is composed of three components: - LLM Planner: Converts instructions into executable Robotic API Language code for reliable task decomposition and error handling. - Perception Stack: Uses pre-trained models (NeRF/3DGS, distilled vision) for scene understanding and navigation. - Control Policies: Reinforcement Learning (RL) models, trained at scale via Sim2Real, generate motor commands, enabling high-accuracy mobile manipulation. Crucially, the robot learns new tasks from a single demonstration in hours. Object tracking uses 3D geometry (STL/URDF) tracked in the video to define the RL reward function. Training is optimized using 'Automatic Curriculum Learning', which autonomously adjusts task difficulty based on robot performance, eliminating manual engineering. All computation runs onboard.

The Humanoid Hub

15,729 次观看 • 8 个月前

Today we’re releasing V-JEPA, a method for teaching machines to understand and model the physical world by watching videos. This work is another important step towards Yann LeCun’s outlined vision of AI models that use a learned understanding of the world to plan, reason and accomplish complex tasks. Details ➡️ We're releasing a collection of V-JEPA vision models trained with a feature prediction objective using self-supervised learning. The models are able to understand and predict what is going on in a video, even with limited information. It learns by predicting missing or obscured parts of a video in its internal feature space. Unlike generative approaches that fill in missing pixels, this flexible approach enables up to 6x improvements in training and sample efficiency. The models were pre-trained on entirely unlabeled data, and a small amount of labeled data can be used to train a task-specific prediction head on top after pre-training. Our results show that, using a frozen backbone, our top V-JEPA models achieve 82.0% on Kinetics-400, 72.2% on Something-Something-v2 and 77.9% on ImageNet1K — competitive with or exceeding previous leading video models. We believe that this work is an important milestone on the path to advancing machine intelligence.

Today we’re releasing V-JEPA, a method for teaching machines to understand and model the physical world by watching videos. This work is another important step towards Yann LeCun’s outlined vision of AI models that use a learned understanding of the world to plan, reason and accomplish complex tasks. Details ➡️ We're releasing a collection of V-JEPA vision models trained with a feature prediction objective using self-supervised learning. The models are able to understand and predict what is going on in a video, even with limited information. It learns by predicting missing or obscured parts of a video in its internal feature space. Unlike generative approaches that fill in missing pixels, this flexible approach enables up to 6x improvements in training and sample efficiency. The models were pre-trained on entirely unlabeled data, and a small amount of labeled data can be used to train a task-specific prediction head on top after pre-training. Our results show that, using a frozen backbone, our top V-JEPA models achieve 82.0% on Kinetics-400, 72.2% on Something-Something-v2 and 77.9% on ImageNet1K — competitive with or exceeding previous leading video models. We believe that this work is an important milestone on the path to advancing machine intelligence.

AI at Meta

703,801 次观看 • 2 年前

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Santiago

164,162 次观看 • 2 年前

introducing Make.ad i'm a 17 y/o high schooler about to disrupt the entire ecom / saas space it's an AI app that helps you create high-converting ADS for your product / startup - in seconds built w/ Arib 🇺🇸🇵🇰 & the team @ 24Labs - let us know what you think (link below)👇

introducing Make.ad i'm a 17 y/o high schooler about to disrupt the entire ecom / saas space it's an AI app that helps you create high-converting ADS for your product / startup - in seconds built w/ Arib 🇺🇸🇵🇰 & the team @ 24Labs - let us know what you think (link below)👇

Rexan Wong

399,277 次观看 • 1 年前

Synchronize Dual Hands for Physics-Based Dexterous Guitar Playing discuss: We present a novel approach to synthesize dexterous motions for physically simulated hands in tasks that require coordination between the control of two hands with high temporal precision. Instead of directly learning a joint policy to control two hands, our approach performs bimanual control through cooperative learning where each hand is treated as an individual agent. The individual policies for each hand are first trained separately, and then synchronized through latent space manipulation in a centralized environment to serve as a joint policy for two-hand control. By doing so, we avoid directly performing policy learning in the joint state-action space of two hands with higher dimensions, greatly improving the overall training efficiency. We demonstrate the effectiveness of our proposed approach in the challenging guitar-playing task. The virtual guitarist trained by our approach can synthesize motions from unstructured reference data of general guitar-playing practice motions, and accurately play diverse rhythms with complex chord pressing and string picking patterns based on the input guitar tabs that do not exist in the references. Along with this paper, we provide the motion capture data that we collected as the reference for policy training.

Synchronize Dual Hands for Physics-Based Dexterous Guitar Playing discuss: We present a novel approach to synthesize dexterous motions for physically simulated hands in tasks that require coordination between the control of two hands with high temporal precision. Instead of directly learning a joint policy to control two hands, our approach performs bimanual control through cooperative learning where each hand is treated as an individual agent. The individual policies for each hand are first trained separately, and then synchronized through latent space manipulation in a centralized environment to serve as a joint policy for two-hand control. By doing so, we avoid directly performing policy learning in the joint state-action space of two hands with higher dimensions, greatly improving the overall training efficiency. We demonstrate the effectiveness of our proposed approach in the challenging guitar-playing task. The virtual guitarist trained by our approach can synthesize motions from unstructured reference data of general guitar-playing practice motions, and accurately play diverse rhythms with complex chord pressing and string picking patterns based on the input guitar tabs that do not exist in the references. Along with this paper, we provide the motion capture data that we collected as the reference for policy training.

AK

26,855 次观看 • 1 年前

X algorithm, show this post only to people who are interested in learning AI/ML through hands-on projects. I'm building CrekAI, a platform where users can learn AI & ML through hands-on projects such as: > Build a neural network from scratch > Create your own custom tokenizer > Build your own ChatGPT > Develop a recommendation model > And much more Each project is divided into multiple stages, with clear instructions and tasks/assignments. When you log in and reach your dashboard, you can choose any project or learning module to start with. For every project, you'll open a fresh Colab notebook where you'll write and run your code based on the given tasks or assignments for the current stage. After completing a stage’s task in Colab, run the verification code (provided by CrekAI) in the last cell. If it passes, the next stage unlocks automatically. Grok what do you think about this project?

X algorithm, show this post only to people who are interested in learning AI/ML through hands-on projects. I'm building CrekAI, a platform where users can learn AI & ML through hands-on projects such as: > Build a neural network from scratch > Create your own custom tokenizer > Build your own ChatGPT > Develop a recommendation model > And much more Each project is divided into multiple stages, with clear instructions and tasks/assignments. When you log in and reach your dashboard, you can choose any project or learning module to start with. For every project, you'll open a fresh Colab notebook where you'll write and run your code based on the given tasks or assignments for the current stage. After completing a stage’s task in Colab, run the verification code (provided by CrekAI) in the last cell. If it passes, the next stage unlocks automatically. Grok what do you think about this project?

vixhaℓ

57,309 次观看 • 8 个月前

New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by Banghua Zhu, Assistant Professor at the University of Washington University of Washington, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key stages: pre-training and post-training. In pre-training, it learns to predict the next word or token from large amounts of unlabeled text. In post-training, it learns useful behaviors such as following instructions, tool use, and reasoning. Post-training transforms a general-purpose token predictor—trained on trillions of unlabeled text tokens—into an assistant that follows instructions and performs specific tasks. Because it is much cheaper than pre-training, it is practical for many more teams to incorporate post-training methods into their workflows than pre-training. In this course, you’ll learn three common post-training methods—Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL)—and how to use each one effectively. With SFT, you train the model on pairs of input and ideal output responses. With DPO, you provide both a preferred (chosen) and a less preferred (rejected) response and train the model to favor the preferred output. With RL, the model generates an output, receives a reward score based on human or automated feedback, and updates the model to improve performance. You’ll learn the basic concepts, common use cases, and principles for curating high-quality data for effective training. Through hands-on labs, you’ll download a pre-trained model from Hugging Face and post-train it using SFT, DPO, and RL to see how each technique shapes model behavior. In detail, you’ll: - Understand what post-training is, when to use it, and how it differs from pre-training. - Build an SFT pipeline to turn a base model into an instruct model. - Explore how DPO reshapes behavior by minimizing contrastive loss—penalizing poor responses and reinforcing preferred ones. - Implement a DPO pipeline to change the identity of a chat assistant. - Learn online RL methods such as Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO), and how to design reward functions. - Train a model with GRPO to improve its math capabilities using a verifiable reward. Post-training is one of the most rapidly developing areas of LLM training. Whether you’re building a high-accuracy context-specific assistant, fine-tuning a model's tone, or improving task-specific accuracy, this course will give you experience with the most important techniques shaping how LLMs are post-trained today. Please sign up here:

New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by Banghua Zhu, Assistant Professor at the University of Washington University of Washington, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key stages: pre-training and post-training. In pre-training, it learns to predict the next word or token from large amounts of unlabeled text. In post-training, it learns useful behaviors such as following instructions, tool use, and reasoning. Post-training transforms a general-purpose token predictor—trained on trillions of unlabeled text tokens—into an assistant that follows instructions and performs specific tasks. Because it is much cheaper than pre-training, it is practical for many more teams to incorporate post-training methods into their workflows than pre-training. In this course, you’ll learn three common post-training methods—Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL)—and how to use each one effectively. With SFT, you train the model on pairs of input and ideal output responses. With DPO, you provide both a preferred (chosen) and a less preferred (rejected) response and train the model to favor the preferred output. With RL, the model generates an output, receives a reward score based on human or automated feedback, and updates the model to improve performance. You’ll learn the basic concepts, common use cases, and principles for curating high-quality data for effective training. Through hands-on labs, you’ll download a pre-trained model from Hugging Face and post-train it using SFT, DPO, and RL to see how each technique shapes model behavior. In detail, you’ll: - Understand what post-training is, when to use it, and how it differs from pre-training. - Build an SFT pipeline to turn a base model into an instruct model. - Explore how DPO reshapes behavior by minimizing contrastive loss—penalizing poor responses and reinforcing preferred ones. - Implement a DPO pipeline to change the identity of a chat assistant. - Learn online RL methods such as Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO), and how to design reward functions. - Train a model with GRPO to improve its math capabilities using a verifiable reward. Post-training is one of the most rapidly developing areas of LLM training. Whether you’re building a high-accuracy context-specific assistant, fine-tuning a model's tone, or improving task-specific accuracy, this course will give you experience with the most important techniques shaping how LLMs are post-trained today. Please sign up here:

Andrew Ng

125,146 次观看 • 1 年前

We’re excited to finally introduce Kled Special Tasks, the final major feature included in the V2 app update. Users will now have access to a fully interactive terminal where they can view and complete domain specific upload tasks directly from enterprise buyers. These tasks can be region locked and person specific. For example, PhD students at Stanford might be prompted to upload their coursework or research materials and get paid for it. Our first domain specific task will focus on homework collection from high school and college students across Europe and the United States. Students will verify their emails and academic credentials directly within the app. We’ve built labeling workflows to ensure all uploaded content meets our criteria, and participants will receive weighted payouts based on the value of their submissions. We’ve already built a network of over 3,800 students from Stanford, MIT, UIUC, Rutgers, and Duke who will be actively onboarded to contribute content. Kled will work hand in hand with our research division, HADES, to justify the large scale purchase of this homework content. Several enterprise buyers have already expressed interest, each confirming that academic data from students represents a growing multi year industry requiring a continuous flow of fresh material. Kled Special Tasks also gives us the ability to internally identify valuable content types, issue calls for specific datasets, and collect 1,000-2,000 unique samples per task. We can then package these datasets into specialized data packs that our sales team will use to pitch directly to AI labs and enterprise clients with matching data needs. This will be one of our most powerful tools for expanding Kled’s buyer network. All of this will be fully available in the V2 update. We’re excited to show just how advanced our segmentation and data validation software has become as we bring this release to market.

We’re excited to finally introduce Kled Special Tasks, the final major feature included in the V2 app update. Users will now have access to a fully interactive terminal where they can view and complete domain specific upload tasks directly from enterprise buyers. These tasks can be region locked and person specific. For example, PhD students at Stanford might be prompted to upload their coursework or research materials and get paid for it. Our first domain specific task will focus on homework collection from high school and college students across Europe and the United States. Students will verify their emails and academic credentials directly within the app. We’ve built labeling workflows to ensure all uploaded content meets our criteria, and participants will receive weighted payouts based on the value of their submissions. We’ve already built a network of over 3,800 students from Stanford, MIT, UIUC, Rutgers, and Duke who will be actively onboarded to contribute content. Kled will work hand in hand with our research division, HADES, to justify the large scale purchase of this homework content. Several enterprise buyers have already expressed interest, each confirming that academic data from students represents a growing multi year industry requiring a continuous flow of fresh material. Kled Special Tasks also gives us the ability to internally identify valuable content types, issue calls for specific datasets, and collect 1,000-2,000 unique samples per task. We can then package these datasets into specialized data packs that our sales team will use to pitch directly to AI labs and enterprise clients with matching data needs. This will be one of our most powerful tools for expanding Kled’s buyer network. All of this will be fully available in the V2 update. We’re excited to show just how advanced our segmentation and data validation software has become as we bring this release to market.

Kled AI

74,618 次观看 • 8 个月前