LeRobot v0.6.0: Imagine, Evaluate, Improve
We're on a journey to advance and democratize artificial intelligence through open source and open science. It also brings depth sensing, VLM-powered dataset annotation, custom video encoding, cloud training on HF Jobs, and a much leaner install. TL;DR LeRobot v0.6.0 introduces world model policies (VLA-JEPA, FastWAM, LingBot-VA) that learn to imagine the future, a wave of new VLAs (GR00T N1.7, MolmoAct2, EO-1, EVO1, Multitask DiT), and a new reward models API (Robometer, TOPReward).
Key Takeaways
- It ships six new simulation benchmarks unified under , the CLI with DAgger-style human-in-the-loop corrections, FSDP training, and cloud training on HF Jobs.
Datasets get depth support, an automatic language annotation pipeline, custom video encoding, and up to 2x faster data loading, all on top of a leaner installation.
- Each one learns to imagine the future as part of its training, and each takes a different path to keep that imagination affordable.
VLA-JEPA VLA-JEPA teaches a compact VLA (built on Qwen3-VL-2B) to predict the future in latent space while it learns to act: during training, a JEPA world model has to anticipate upcoming frames from the model's own actions.
- LingBot-VA LingBot-VA goes one step further: an autoregressive video-action model that predicts future video and actions together, chunk by chunk, and feeds real observations back in to keep its imagination grounded.
You can even save what the robot imagined ( ) and compare it with what actually happened.
- FastWAM FastWAM asks the question in its paper title: do world action models need test-time future imagination?
It pairs a ~5B video-generation expert with a compact action expert in a single network, so the model literally learns to dream its own rollouts.
- VLAs: the model zoo keeps growing GR00T N1.7 We upgraded our NVIDIA GR00T integration to GR00T N1.7, the newest open generation of NVIDIA's cross-embodiment foundation model.
Stats & Key Facts
- #Datasets get depth support, an automatic language annotation pipeline, custom video encoding, and up to 2x faster data loading, all on top of a leaner installation.
It ships six new simulation benchmarks unified under , the CLI with DAgger-style human-in-the-loop corrections, FSDP training, and cloud training on HF Jobs. Datasets get depth support, an automatic language annotation pipeline, custom video encoding, and up to 2x faster data loading, all on top of a leaner installation. v0.6.0 brings three policies to LeRobot to help answer that question.
Each one learns to imagine the future as part of its training, and each takes a different path to keep that imagination affordable. VLA-JEPA VLA-JEPA teaches a compact VLA (built on Qwen3-VL-2B) to predict the future in latent space while it learns to act: during training, a JEPA world model has to anticipate upcoming frames from the model's own actions. The trick is that the world model then disappears at inference, so you get world-model supervision at zero extra inference cost.
Three ready-to-use checkpoints are on the Hub, including a DROID-pretrained base for fine-tuning: Check out the VLA-JEPA documentation and the paper to learn more. LingBot-VA LingBot-VA goes one step further: an autoregressive video-action model that predicts future video and actions together, chunk by chunk, and feeds real observations back in to keep its imagination grounded. You can even save what the robot imagined ( ) and compare it with what actually happened.
For more details please read the original article at Hugging Face.
Continue Learning
Comments
Comments appear only after moderation. Your email identifies your submission to the moderator and is never displayed here.
No approved comments yet.