Skip to content
Robotics

SmolVLA SO101

политика сортировки кубиков

1 min readupdated

Это файнтюн компактной vision-language-action модели SmolVLA от автора hubnemo для робота so101_follower. Политика обучена в LeRobot брать фиолетовый кубик и класть его в картонную коробку, игнорируя другие объекты. База — lerobot/smolvla_base, лицензия apache-2.0.

Ключевые особенности

  • Вход: состояние робота (6,) и одно изображение с камеры front 3×480×640.
  • Выход: вектор действий (6,).
  • Обучение на датасете so101_sort_cubes_no_top: 77 эпизодов, 54 777 кадров, 30 FPS.
  • 10 000 шагов, batch size 16, AdamW, learning rate 1e-4, seed 1000, LeRobot 0.7.1.
  • Запуск и дообучение через CLI LeRobot; по описанию SmolVLA модель можно разворачивать на consumer-grade железо.

Пример

Запуск политики на роботе:

lerobot-rollout \
  --strategy.type=base \
  --robot.type=so101_follower \
  --robot.port=<your_robot_port> \
  --robot.cameras="{ front: {type: opencv, index_or_path: <index>, width: 640, height: 480, fps: 30} }" \
  --policy.path=hubnemo/so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4 \
  --task="Grab the purple cube and put it into the cardboard box, ignore the other objects" \
  --duration=60
ПодробнееОценки на реальном роботе в карточке нет.
Данные: 77 эпизодов, 54 777 кадров при 30 FPS.
Обучение: 10k шагов, batch 16, lr 1e-4, AdamW, seed 1000, LeRobot 0.7.1.
Для запуска нужны порт робота и камеры, совпадающие с observation-ключами; лицензия apache-2.0.
Discuss Telegram 0 23

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.9 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.

This is a LoRA: counted together with its base model smolvla_base.

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.

    # hubnemo/so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4
    # ===== 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/smolvla_base --local-dir ~/models/smolvla_base
    hf download hubnemo/so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4 --local-dir ~/models/so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to smolvla_base
  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/smolvla_base

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

  1. SmolLM2-360Mfamily root
  2. SmolLM2-360M-Instruct
  3. SmolVLM-500M-Instruct
  4. SmolVLM2-500M-Video-Instruct
  5. smolvla_base
  6. so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4

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

hubnemo/so101_smolvla_exp02_fixed_bs16_steps10k_lr5e-4
Author
hubnemo
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

Robotics

vla_ma

vla_ma