Skip to content
Robotics

vla_ma

LoRA-адаптер для роботизированных манипуляций

1 min readupdated

vla_ma — LoRA-адаптер для FineART-VLA от pmolodyk. Он восстанавливает адаптер, процессоры и токенизаторы на закреплённой базовой модели lerobot/FineART-VLA-subtask-midtrained. Предназначен для роботизированных манипуляций: по состоянию и трём RGB-камерам выбирает следующую подзадачу и предсказывает последовательность действий.

Ключевые особенности

  • LoRA-адаптер поверх FineART-VLA; загрузчик возвращает policy, preprocessor и postprocessor.
  • На входе 14-элементный вектор состояния и три RGB-изображения: cam_high, cam_left_wrist, cam_right_wrist.
  • Планировщик выбирает следующую подзадачу через select_message, затем predict_action_chunk выдаёт 50 шагов действий.
  • Выход — absolute action targets формы [1, 50, 14].
  • Проверено с LeRobot commit 72e1d9e5e76ee719bacfc78bc34d817bee5fb521, Transformers 5.5.4 и PEFT 0.21.2; нужен Python 3.12 и CUDA-PyTorch.

Пример

import sys
from huggingface_hub import snapshot_download
root = snapshot_download("pmolodyk/vla_ma")
sys.path.insert(0, root)
from load_policy import load_policy
policy, preprocessor, postprocessor = load_policy(root, device="cuda")
# далее: obs + task -> select_message -> predict_action_chunk
ПодробнееБаза: lerobot/FineART-VLA-subtask-midtrained, ревизия b6cf4faf2630f971dd390bc32365e9d9c330da8e.
Бенчмарков и требований к VRAM в карточке нет; указаны Python 3.12, CUDA-PyTorch и фиксированные версии LeRobot/Transformers/PEFT.
Порядок суставов — в config.json; для инференса нужны три RGB-тензора [0,1] и 14-мерный state. Лицензия наследует условия Gemma.
Discuss Telegram 0 23

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 8.7 GB · ready files10 GB
  • FP8 / INT8recommendedweights 3.9 GB4.8 GB
  • 4-bitweights 2.2 GB2.9 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 2080 Ti

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

This is a LoRA: counted together with its base model FineART-VLA-subtask-midtrained.

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.

    # pmolodyk/vla_ma
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated 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 lerobot/FineART-VLA-subtask-midtrained --local-dir ~/models/FineART-VLA-subtask-midtrained
    hf download pmolodyk/vla_ma --local-dir ~/models/vla_ma
    # 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

Requires the base model

lerobot/FineART-VLA-subtask-midtrained

This model is a LoRA / adapter. Download the base model first, then load this one on top.

  1. paligemma-3b-pt-224family root
  2. pi05_base
  3. FineART-VLA-subtask-midtrained
  4. vla_ma

New posts about versions of paligemma-3b-pt-224 (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 12All on Hugging Face
LoRAs and adapters 12All on Hugging Face

About the model

pmolodyk/vla_ma
Author
pmolodyk
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

230

Ask AI about this model

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

Comments