Skip to content
Robotics

dp-blockpick-2k-withoutpush-cotrain-real075

diffusion policy для захвата блока

1 min read

Это LeRobot diffusion policy для робота ROBOTIS FFW SG2: левая рука сверху захватывает лаймово-зелёный блок. Модель от padiredd из форка ROBOTIS lerobot-cyclo, 288 M параметров; на входе три камеры и joint state, на выходе 16 joint targets с частотой 15 Hz.

Ключевые особенности

  • Обучение на 2 014 эпизодах: 2 000 sim-демо и 14 реальных teleop-демо; co-training с alpha 0.75 в пользу реальных данных.
  • Входы: камеры head, left wrist и right wrist по 240×240 плюс 22-мерный joint state.
  • Рецепт: 100k steps, batch 32, ResNet18 на каждую камеру, Adam 1e-4, cosine, DDPM 100 steps.
  • Горизонт 64, исполняется 32 шага; n_obs_steps=1; root-чекпоинт соответствует step 100k, промежуточные — каждые 10k.
  • Аугментации изображений тяжёлые: яркость, контраст, saturation, hue, sharpness, affine, perspective, blur, random erasing, grayscale.

Пример

policy = DiffusionPolicy.from_pretrained("padiredd/dp-blockpick-2k-withoutpush-cotrain-real075")
ПодробнееПодробнее
Обучение: 100k steps, batch 32, ResNet18 на камеру, Adam 1e-4, cosine, DDPM 100 steps, horizon 64 / 32 executed.
Железо: одна NVIDIA A40 46 GB, ~5.7 steps/s.
Ограничения: задача строго «pick up the lime green block with the left arm»; результаты оценки в карточке пока не добавлены.
Discuss Telegram 0 26

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 1.1 GB · ready files1.7 GB
  • BF16weights 0.5 GB1.1 GB
  • FP8 / INT8recommendedweights 0.3 GB0.8 GB
  • 4-bitweights 0.2 GB0.7 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 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.

    # padiredd/dp-blockpick-2k-withoutpush-cotrain-real075
    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 padiredd/dp-blockpick-2k-withoutpush-cotrain-real075 --local-dir ~/models/dp-blockpick-2k-withoutpush-cotrain-real075
    # 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 dp-blockpick-2k-withoutpush-cotrain-real075 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

padiredd/dp-blockpick-2k-withoutpush-cotrain-real075
Author
padiredd
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

260

Ask AI about this model

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

Comments