Skip to content
Robotics

GR00T-политика для SO-101

перестановка флаконов по команде

1 min readupdated

Это обученная в LeRobot политика на базе GR00T N1.7 — открытой кросс-эмбодимент foundation-модели NVIDIA для гуманоидных роботов. Она управляет so_follower по двум камерам (overhead, wrist) и 6-мерному состоянию, а на выходе даёт 6-мерный вектор действий. Задача — брать белый или синий флакон и ставить его в одно из четырёх отверстий справа.

Ключевые особенности

  • 160 эпизодов и 95 178 кадров при 30 FPS; 8 формулировок задач с разным цветом флакона и отверстием.
  • Входы: observation.state (6,), две камеры 480×640.
  • Выход: action (6,) для робота so_follower.
  • Обучение: 20 000 шагов, batch 32, AdamW, lr 0.0001, seed 42, LeRobot 0.6.1.
  • Лицензия Apache-2.0; оценка на реальном роботе в карточке пока не опубликована.

Пример

lerobot-rollout --strategy.type=base --robot.type=so_follower --robot.port=<port> --robot.cameras='{ overhead: {type: opencv, index_or_path: <idx1>, width: 640, height: 480, fps: 30}, wrist: {type: opencv, index_or_path: <idx2>, width: 640, height: 480, fps: 30}}' --policy.path=yudai0209/so101_vial2_train160_swapaug_groot17 --task='Pick up the white vial and put it in the 1st hole from the right.'
ПодробнееОбучение: 20 000 шагов, batch 32, AdamW, lr 0.0001, seed 42, LeRobot 0.6.1.
Данные: 160 эпизодов, 95 178 кадров, 30 FPS; 8 задач на перестановку флаконов.
Лицензия Apache-2.0. Оценки на реальном роботе и требований к VRAM в карточке нет.
Discuss Telegram 0 13

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 12 GB · ready files13 GB
  • BF16weights 5.9 GB6.9 GB
  • FP8 / INT8recommendedweights 2.9 GB3.7 GB
  • 4-bitweights 1.6 GB2.3 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 · F32 — from RTX 5080

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.

    # yudai0209/so101_vial2_train160_swapaug_groot17
    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 yudai0209/so101_vial2_train160_swapaug_groot17 --local-dir ~/models/so101_vial2_train160_swapaug_groot17
    # The model is in ~/models. How to run it is in the model card (link above).
  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 so101_vial2_train160_swapaug_groot17 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

yudai0209/so101_vial2_train160_swapaug_groot17
Author
yudai0209
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

130

Ask AI about this model

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

Comments