TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

robotics

A Humanoid Model Split Action Tokens by Body Part

Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.

Published Updated Story ID: mp-2026-09-29-009
Read the complete editionStory JSON

Summary

Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.

Holo-M extends a language model's vocabulary with four action-token groups for end effectors, body, hands and kinematics. The design trains across humanoid teleoperation, egocentric human video and simulation, then decodes body-part groups with discrete diffusion rather than token-by-token autoregression. The authors report the highest success rates in both generalist and specialist evaluations on their SIMPLE humanoid benchmark. That ranking belongs to the submitted comparison set; code and weights are promised for release.

Why it matters

Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.

Limits and context

No additional limitation was separately recorded.

Key claims

  1. Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.

    Evidence: source-2026-09-29-009

Sources

  1. arXiv preprint 2609.35709arXiv · primary research

Corrections

No corrections have been recorded for this story.