Skip to content
Robotics

GoedelMachines/go2-world-models

Checkpoints for quad-pedal robot motion forecasting

2 min read

A collection of 25 checkpoints in FP32 format for predicting future sensor readings and joint states of the Unitree Go2 robot 30 steps ahead at a 50 Hz frequency. The models operate as open-loop motion forecasting systems, utilizing historical measurements and target joint positions. The root weights require terrain scanning, whereas the adapted models from the proprio folder function without scan data. This set is useful for engineers and researchers working in robotics and quadrupedal motion simulation.

Key features

  • 25 FP32 checkpoints for forecasting Unitree Go2 robot states at 50 Hz
  • Root models with approximately 120M parameters and terrain scanning support
  • Adapted flagship model with 119.6M parameters for operation without scans
  • Availability of configurations, manifests, and equivalent exports in SafeTensors format

What you can do

  • Apply the models to forecast trajectories and joint states for the Unitree Go2 robot
  • Test motion simulations based on measured history and target joints
  • Evaluate forecast accuracy on flat ground and real robot SDK recordings

Example

Use the manifest and configuration files from the repository along with the companion project to load weights and run examples.

Read moreDetails
Benchmarks: the scanned model achieves 0.364 normalized endpoint error over 3 seconds; the adapted model achieves 0.613 over 1 s and 1.009 over 3 s.
Requirements: PyTorch and SafeTensors for loading weights.
Limitations: positions, linear velocity, and foot contacts are excluded from the output interface.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.4 GB · ready files1.0 GB
  • FP8 / INT8recommendedweights 0.2 GB0.7 GB
  • 4-bitweights 0.1 GB0.6 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # GoedelMachines/go2-world-models
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download GoedelMachines/go2-world-models --local-dir ~/models/go2-world-models
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to go2-world-models
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of go2-world-models (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

GoedelMachines/go2-world-models
Author
GoedelMachines
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments