Skip to content
Robotics

smolvla_so101_left_arm_3tasks

компактная VLA-политика для трёх задач

2 min read

Это файнтюн lerobot/smolvla_base на базе SmolVLA — компактной vision-language-action модели (450M параметров), обученной для робота so_follower. Политика управляет левой рукой SO-101 по изображениям с камер top и wrist и текстовой инструкции; в обучении — три манипуляции с кубиками и ящиком.

Ключевые особенности

  • Понимает текстовую инструкцию: подмена задачи заметно меняет предсказанные действия.
  • Компактная: 450M параметров, обучался только action expert (~100M), vision encoder заморожен.
  • Три задачи: достать красный куб из кастрюли, открыть ящик и положить блок, сложить куб на синий.
  • Запуск и дообучение — через CLI LeRobot; на rollout камеры должны называться top и wrist.
  • Все 159 эпизодов попали в train: open-loop метрики — не оценка обобщения и не success rate.

Пример

lerobot-rollout \
  --strategy.type=base \
  --robot.type=so_follower \
  --robot.port=<your_robot_port> \
  --robot.cameras="{ top: {type: opencv, index_or_path: <idx>, width: 640, height: 360, fps: 30}, wrist: {type: opencv, index_or_path: <idx>, width: 640, height: 480, fps: 30}}" \
  --policy.path=khanhnd61/smolvla_so101_left_arm_3tasks \
  --task="Take the red cube out of the steel pot and place it into the orange hollow cube." \
  --duration=60
ПодробнееOpen-loop MAE 1.66 против 8.64 у baseline «держать текущее состояние»; при подмене инструкции — 6.93.
Обучение: 41 801 шаг, batch 32, ~15 ч 17 мин на RTX 3060 12 GB; финальный training loss 0.024.
Реальных испытаний на роботе нет; предсказанные чанки примерно в 7 раз «нервнее» демонстраций — поможет сглаживание действий.
Примеры
Discuss Telegram 0 12

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.8 GB · ready files1.4 GB
  • FP8 / INT8recommendedweights 0.4 GB1.0 GB
  • 4-bitweights 0.2 GB0.8 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.

    # khanhnd61/smolvla_so101_left_arm_3tasks
    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 khanhnd61/smolvla_so101_left_arm_3tasks --local-dir ~/models/smolvla_so101_left_arm_3tasks
    # 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

  1. SmolLM2-360Mfamily root
  2. SmolLM2-360M-Instruct
  3. SmolVLM-500M-Instruct
  4. SmolVLM2-500M-Video-Instruct
  5. smolvla_base
  6. smolvla_so101_left_arm_3tasks

New posts about versions of SmolLM2-360M (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 8All on Hugging Face
Merges 1All on Hugging Face

About the model

khanhnd61/smolvla_so101_left_arm_3tasks
Author
khanhnd61
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

120

Ask AI about this model

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

Comments

Robotics

ACT Jenga2

ACT Jenga2