Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

⭐ The first foundational model available on LeRobot ⭐ Pi0 is the most advanced Vision Language Action model. It takes natural language commands as input and directly output autonomous behavior. It was trained by Physical Intelligence and ported to pytorch by Pablo Montalvo 👇🧵

RemiCadene's profile picture

Remi Cadene

20,749 subscribers

130,941 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

11 Kommentare

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

Take a look at the code:

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

You can now download and use the pretrained policy just like transformers: policy = Pi0Policy.from_pretrained("lerobot/pi0") action =

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

Even cooler, you can finetune it directly on your dataset. Here is an example with @DAubakirovaa 's dataset: python lerobot/scripts/train.py \ --policy.path=lerobot/pi0 \ --dataset.repo_id=danaaubakirova/koch_test Visualize her dataset here:

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

The original code in Jax is available over here:

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

We have full blog post with more details!

Profilbild von AssemblyAI
AssemblyAIvor 1 Jahr

Announcing: Our most advanced speech-to-text model goes beyond accuracy to capture the real-world complexity of human conversation and deliver reliable, source-of-truth audio data. Explore Universal-2 updates 👇

Profilbild von Muhammad Ahmed
Muhammad Ahmedvor 1 Jahr

@LeRobotHF @physical_int @m_olbap that is so cool , Thanks @LeRobotHF ! Is this model somewhat similar to OpenVLA (which is also on hugging face) ? For instance with OpenVLA we get action output tokens for a single armed robot. Can i simply replace OpenVLA with this without changing the inference pipeline ?

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

@LeRobotHF @physical_int @m_olbap Pi0 was trained with much more data! Yes you should be as our policy includes all the preprocessing and post processing. Only requirements: images need to be in float32 [0,1] range.

Profilbild von Igor Beaver
Igor Beavervor 1 Jahr

@LeRobotHF @physical_int @m_olbap Incredible gift!

Profilbild von Venky
Venkyvor 1 Jahr

@LeRobotHF @physical_int @m_olbap Super. The LLM moment for Robotics is here. The gap is closing fast. Eagerly awaiting for the weights. Also, would be great to have some blog or tutorial on how one can do this from scratch.

Profilbild von Remi Cadene
Remi Cadenevor 1 Jahr

@LeRobotHF @physical_int @m_olbap We ported the weights already. We are working on a tutorial to reproduce this on more affordable robots

Ähnliche Videos

📢 First contact between a frontier model and robots! Gemini Robotics is a SOTA generalist Vision-Language-Action model bringing frontier model intelligence to the physical world. It's an extremely capable model enabling dexterous, steerable, and general robot control. 🧵⬇️
0:31

📢 First contact between a frontier model and robots! Gemini Robotics is a SOTA generalist Vision-Language-Action model bringing frontier model intelligence to the physical world. It's an extremely capable model enabling dexterous, steerable, and general robot control. 🧵⬇️

Ted Xiao

152,458 Aufrufe • vor 1 Jahr

Vision RAG with vector database is all you need. It uses vision language model to embed pages of PDF as directly vectors, without the tedious chunking process. 100% Opensource code.
0:38

Vision RAG with vector database is all you need. It uses vision language model to embed pages of PDF as directly vectors, without the tedious chunking process. 100% Opensource code.

Shubham Saboo

106,785 Aufrufe • vor 1 Jahr

We’re bringing powerful AI directly onto robots with Gemini Robotics On-Device. 🤖 It’s our first vision-language-action model to help make robots faster, highly efficient, and adaptable to new tasks and environments - without needing a constant internet connection. 🧵
0:35

We’re bringing powerful AI directly onto robots with Gemini Robotics On-Device. 🤖 It’s our first vision-language-action model to help make robots faster, highly efficient, and adaptable to new tasks and environments - without needing a constant internet connection. 🧵

Google DeepMind

820,024 Aufrufe • vor 1 Jahr

🕹️ Is this the smallest language model in the world? I just managed to squeeze JAM, real artificial intelligence into 30 kilobytes, running on a 1979 Atari 800. Just A Model. Fully generative, deterministic language model, powered by a neural network and built to run on 8‑bit hardware. Atari Forever.
2:13

🕹️ Is this the smallest language model in the world? I just managed to squeeze JAM, real artificial intelligence into 30 kilobytes, running on a 1979 Atari 800. Just A Model. Fully generative, deterministic language model, powered by a neural network and built to run on 8‑bit hardware. Atari Forever.

Marek Spanel

113,782 Aufrufe • vor 5 Monaten

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):
0:39

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Léo

32,331 Aufrufe • vor 2 Monaten

Exciting progress on Vision-Language-Action models from a collaboration between San Francisco-based Physical Intelligence (π) and China’s AGIBOT: (π)’s single model can autonomously perform diverse tasks on the AGIBOT G1 robot, using both humanoid hands and two-finger grippers.
3:12

Exciting progress on Vision-Language-Action models from a collaboration between San Francisco-based Physical Intelligence (π) and China’s AGIBOT: (π)’s single model can autonomously perform diverse tasks on the AGIBOT G1 robot, using both humanoid hands and two-finger grippers.

The Humanoid Hub

32,593 Aufrufe • vor 1 Jahr

Below is our open-source arm assembled by one of the discord members! Also, today Physical Intelligence shared their incredible top-notch robotics model Pi0. And I just want to remind you that you can use our arm together with this model and with LeRobot The arm has the same DOF as the arm used by Pi0, but has smaller size and cost starts from 200$ per arm. You can use better motors like we did in our demos, then it will be more precise and stable but the price will be about $600 per arm. Printable files, purchase list, and help with the assembly - in our discord channel.
0:25

Below is our open-source arm assembled by one of the discord members! Also, today Physical Intelligence shared their incredible top-notch robotics model Pi0. And I just want to remind you that you can use our arm together with this model and with LeRobot The arm has the same DOF as the arm used by Pi0, but has smaller size and cost starts from 200$ per arm. You can use better motors like we did in our demos, then it will be more precise and stable but the price will be about $600 per arm. Printable files, purchase list, and help with the assembly - in our discord channel.

Igor Kulakov

53,711 Aufrufe • vor 1 Jahr

VLA-JEPA just dropped in LeRobot 🤖 What makes this model special is that it does not just learn what action to take from a given observation, it also leverages a JEPA world model to learn action-relevant dynamics. During training, the VLA leverages V-JEPA2 by conditioning its predictor. This clever trick adds a world modeling objective to the training, which also allows pretraining on human videos. At inference, the world model is dropped entirely, keeping only a standard VLA architecture: Qwen backbone and action head. The demo here was only fine-tuned on 13 examples, showing great pretraining capability and running in real time on NVIDIA Robotics DGX Spark! VLA-JEPA is the first world model to be ported to LeRobot, and I feel like it won't be the last 🚀 Thomas Wolf clem 🤗
0:35

VLA-JEPA just dropped in LeRobot 🤖 What makes this model special is that it does not just learn what action to take from a given observation, it also leverages a JEPA world model to learn action-relevant dynamics. During training, the VLA leverages V-JEPA2 by conditioning its predictor. This clever trick adds a world modeling objective to the training, which also allows pretraining on human videos. At inference, the world model is dropped entirely, keeping only a standard VLA architecture: Qwen backbone and action head. The demo here was only fine-tuned on 13 examples, showing great pretraining capability and running in real time on NVIDIA Robotics DGX Spark! VLA-JEPA is the first world model to be ported to LeRobot, and I feel like it won't be the last 🚀 Thomas Wolf clem 🤗

LeRobot

320,916 Aufrufe • vor 3 Monaten

🚨 BREAKING: Microsoft's first robotics foundation model! 🤯 Microsoft just announced Rho-alpha (ρα), their first robotics model derived from the Phi series of vision-language models. Rho-alpha translates natural language commands into control signals for robotic systems performing bimanual manipulation tasks. Commands like "push the green button with the right gripper," "pull out the red wire," "flip the top switch on," or "turn the knob to position 5" get executed directly by dual-arm robots. What makes this different from standard vision-language-action (VLA) models is the additional modalities. Rho-alpha is a VLA+ model that adds tactile sensing to the perceptual mix, with plans to incorporate force feedback. On the learning side, the model is designed to continually improve during deployment by learning from human feedback. The training approach combines trajectories from physical demonstrations and simulated tasks with web-scale visual question answering data. Since teleoperation data is scarce and expensive, Microsoft is using NVIDIA Isaac Sim on Azure to generate physically accurate synthetic datasets via reinforcement learning. These simulated trajectories get combined with commercial and open physical demonstration datasets. The model is currently under evaluation on dual-arm setups and humanoid robots. Microsoft is opening an Early Access Program for organizations interested in evaluating Rho-alpha. Robots that can adapt to dynamic situations and human preferences are more useful in real environments and more trusted by the people operating them. Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:34

🚨 BREAKING: Microsoft's first robotics foundation model! 🤯 Microsoft just announced Rho-alpha (ρα), their first robotics model derived from the Phi series of vision-language models. Rho-alpha translates natural language commands into control signals for robotic systems performing bimanual manipulation tasks. Commands like "push the green button with the right gripper," "pull out the red wire," "flip the top switch on," or "turn the knob to position 5" get executed directly by dual-arm robots. What makes this different from standard vision-language-action (VLA) models is the additional modalities. Rho-alpha is a VLA+ model that adds tactile sensing to the perceptual mix, with plans to incorporate force feedback. On the learning side, the model is designed to continually improve during deployment by learning from human feedback. The training approach combines trajectories from physical demonstrations and simulated tasks with web-scale visual question answering data. Since teleoperation data is scarce and expensive, Microsoft is using NVIDIA Isaac Sim on Azure to generate physically accurate synthetic datasets via reinforcement learning. These simulated trajectories get combined with commercial and open physical demonstration datasets. The model is currently under evaluation on dual-arm setups and humanoid robots. Microsoft is opening an Early Access Program for organizations interested in evaluating Rho-alpha. Robots that can adapt to dynamic situations and human preferences are more useful in real environments and more trusted by the people operating them. Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

61,026 Aufrufe • vor 8 Monaten

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.
0:42

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.

Bidhan Roy

457,494 Aufrufe • vor 3 Monaten

I trained a small model to do syntax highlighting in the browser with GPU. Meet gpu-lexer from Vercel Labs: Small (27.5KB), fast (runs on WebGPU), and language-agnostic (model guesses the syntax). It is experimental and built for learning!
0:49

I trained a small model to do syntax highlighting in the browser with GPU. Meet gpu-lexer from Vercel Labs: Small (27.5KB), fast (runs on WebGPU), and language-agnostic (model guesses the syntax). It is experimental and built for learning!

Shu

790,939 Aufrufe • vor 15 Tagen

Introducing Realtime TTS-2, a new generation of voice model built for realtime conversation. It is the first voice model that hears the conversation, takes natural-language voice direction, holds one voice identity across over 100 languages, and speaks like a person who is paying attention. The result is voice AI that feels as good as it sounds. Try it out: Learn More:
2:06

Introducing Realtime TTS-2, a new generation of voice model built for realtime conversation. It is the first voice model that hears the conversation, takes natural-language voice direction, holds one voice identity across over 100 languages, and speaks like a person who is paying attention. The result is voice AI that feels as good as it sounds. Try it out: Learn More:

Inworld AI

327,303 Aufrufe • vor 4 Monaten

Aging has a language. 🧬 We’re teaching AI to read it. On the cover of Cell: compact Longevity-LLMs, jointly trained by Insilico Medicine and Liquid AI, demonstrate SOTA performance on LongevityBench. Powered by these models, Longevity Claw nominated 328 candidate gene targets for aging research. Small model. Big questions. Open source. 🧵👇 #insilicoSOTAFM
0:32

Aging has a language. 🧬 We’re teaching AI to read it. On the cover of Cell: compact Longevity-LLMs, jointly trained by Insilico Medicine and Liquid AI, demonstrate SOTA performance on LongevityBench. Powered by these models, Longevity Claw nominated 328 candidate gene targets for aging research. Small model. Big questions. Open source. 🧵👇 #insilicoSOTAFM

Insilico Medicine

16,568 Aufrufe • vor 6 Tagen

🚀 MiniCPM enters the physical world — enabling robots to understand, remember, and act. We open-source MiniCPM-Robot, our first embodied AI model series, including: 🤖 MiniCPM-RobotManip — a 1.5B general-purpose Vision-Language-Action (VLA) model for robotic manipulation. 🐕 MiniCPM-RobotTrack — a compact model for real-world target tracking. ⚡ PhyAI — a high-performance inference framework built for embodied models. Together, they bring efficient, practical, and open embodied intelligence closer to real-world robots. ⭐ GitHub: 🤗 MiniCPM-RobotManip: 🤗 MiniCPM-RobotTrack:
1:17

🚀 MiniCPM enters the physical world — enabling robots to understand, remember, and act. We open-source MiniCPM-Robot, our first embodied AI model series, including: 🤖 MiniCPM-RobotManip — a 1.5B general-purpose Vision-Language-Action (VLA) model for robotic manipulation. 🐕 MiniCPM-RobotTrack — a compact model for real-world target tracking. ⚡ PhyAI — a high-performance inference framework built for embodied models. Together, they bring efficient, practical, and open embodied intelligence closer to real-world robots. ⭐ GitHub: 🤗 MiniCPM-RobotManip: 🤗 MiniCPM-RobotTrack:

OpenBMB

330,472 Aufrufe • vor 2 Monaten

I’m excited to announce the world’s first Fly Language Model. It uses a new architecture called GPF (Generative Pre-trained Fly). It adapts a small pre-trained model with the actual fly connectome. Don't believe it's real? Try it out below, view the code, and read the paper. What will you ask the fly?
0:30

I’m excited to announce the world’s first Fly Language Model. It uses a new architecture called GPF (Generative Pre-trained Fly). It adapts a small pre-trained model with the actual fly connectome. Don't believe it's real? Try it out below, view the code, and read the paper. What will you ask the fly?

Alex Wormuth

98,020 Aufrufe • vor 11 Tagen

It’s time to go beyond language models. Introducing Odyssey-2 Max, our most powerful world model yet. It materially advances the SOTA in physical accuracy. This is a big step toward models that simulate and interact with the world in real time. A new intelligence entirely!
1:08

It’s time to go beyond language models. Introducing Odyssey-2 Max, our most powerful world model yet. It materially advances the SOTA in physical accuracy. This is a big step toward models that simulate and interact with the world in real time. A new intelligence entirely!

Odyssey

2,187,202 Aufrufe • vor 5 Monaten

Today we launched Gemini 3.1 Flash TTS, our most expressive and controllable text-to-speech model yet. This launch [excitement] includes audio tags! 🗣🏷 Audio tags [explanatory] are a seamless way to guide vocal style, pace, and delivery using natural language commands embedded directly in your text. Want a different tempo or tone? [amazement] Just tag the audio to steer the AI-speech output! The model supports 70+ languages (24 of which are high-quality evaluated languages, including: Japanese, Hindi, and Arabic). Watch the audio tags in action in the demo below ↓
1:14

Today we launched Gemini 3.1 Flash TTS, our most expressive and controllable text-to-speech model yet. This launch [excitement] includes audio tags! 🗣🏷 Audio tags [explanatory] are a seamless way to guide vocal style, pace, and delivery using natural language commands embedded directly in your text. Want a different tempo or tone? [amazement] Just tag the audio to steer the AI-speech output! The model supports 70+ languages (24 of which are high-quality evaluated languages, including: Japanese, Hindi, and Arabic). Watch the audio tags in action in the demo below ↓

Google AI

203,435 Aufrufe • vor 5 Monaten