mcMotionChallengeGitHub ↗

MODEL / CHECKPOINT

NVIDIA

Model

Checkpoint artifacts and an independent MotionChallenge inference adapter.

Organization
NVIDIA
Type
FSQ
Quantized latent control using finite scalar quantization (FSQ), with a motion encoder and a conditioned decoder.
Version
Low-Latency · 2026-06-16
Author release
Robot support
Unitree G1 · 29 DoF
Author source revision
45c03facbf848e0c743091b7d9c17952096ff908

Checkpoint artifacts

Author release
NVlabs/GR00T-WholeBodyControl ↗
Artifact type
Author-released checkpoint
Original file
model_decoder.onnx ↗

The adapter pins this published artifact by SHA256. Artifact identity is separate from inference parity.

Download links deliver the adapter files. The release links and verification notes distinguish author files from conversions; SHA256 identifies the bytes used.

Policy checkpoint
Download ↗
c4ac2e74045e7cbfb568f15e6bf47ea7ce023df7a94322af50be223e0a628bab
encoder
Download ↗
60be43157f57d812f38bdbb740a5de5d5d070e8840d9edc16f02a91a6d06255b

Inference interface

The adapter declares its control clock and reference window below. Source and configuration describe the observation and action transforms.

Observation requirements

Observation rules ↗

Model-specific handling

Released actor uses joint state, angular velocity/gravity or orientation and reference features. Retain author normalization, history/recurrent state, action processing and declared lookahead; validate numerical independence from actual translation.

Exact checkpoint & evidence ↗
Control interval
20 ms
Reference history
0 ms
Reference preview
180 ms
Output interface
joint-pd-feedforward
Configuration
policy.json

Evaluations