Skip to content
Robotics

π₀.₅ PSM2 Ring Transfer

политика для переноса кольца на dVRK

1 min readupdated

Это файнтюн lerobot/pi05_base (3,62 млрд параметров) от arkojit1, обученный в LeRobot 0.4.4 на датасете wissenschaftderlogik/psm2-ring-transfer-binjaw-v5. В основе — 49 эпизодов телеуправления dVRK PSM2 на 15 Гц с одной задачей: захватить объект на правом стержне, перенести на средний, отпустить и убрать инструмент. Политика выдаёт чанк из 50 шагов (3,3 с) с 10-мерными действиями.

Ключевые особенности

  • Одна задача ring-transfer; observation.state — 8-D: xyz, quaternion xyzw и угол jaw.
  • Входы с камер: observation.image.cam0 (внешняя) и observation.image.cam1 (wrist), плюс empty_cameras=1.
  • Обучался только action expert (train_expert_only=true); SigLIP и Gemma-2B заморожены.
  • 8 × MI300X, global batch 64; чекпоинт step 16 000 из 20 000 (55,4 эпохи).
  • Лучший eval loss в запуске — 0,1403 на 5 отложенных эпизодах; train loss 0,122.

Пример

from lerobot.policies.pi05.modeling_pi05 import PI05Policy

policy = PI05Policy.from_pretrained("arkojit1/pi05_psm2ring_bs64_16k_20k")
ПодробнееEval loss 0,1403 на эпизодах 44–48; это loss, а не task success rate.
Чекпоинт step 16 000 (55,4 эпохи) — лучший по eval loss в 20 000-шаговом запуске.
Нужны LeRobot 0.4.4 и форк transformers для π₀ (ветка fix/lerobot_openpi).
Действия — дельты как в датасете: action[t] = measured[t+1] − measured[t]; применять так же при деплое.
Нормализация quantile q01–q99; у dims 3 и 7 диапазон ~1e-6, поэтому значения уходят до ~±24.
Discuss Telegram 0 20

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

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.

    # arkojit1/pi05_psm2ring_bs64_16k_20k
    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 arkojit1/pi05_psm2ring_bs64_16k_20k --local-dir ~/models/pi05_psm2ring_bs64_16k_20k
    # 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 pi05_psm2ring_bs64_16k_20k (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

arkojit1/pi05_psm2ring_bs64_16k_20k
Author
arkojit1
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

200

Ask AI about this model

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

Comments