Skip to content
Robotics

smolvla-keys101-22keys

VLA model for playing keys

2 min read

A compact VLA (Vision-Language-Action — an AI model that takes an image and text prompt as input and outputs motor actions) robotic model trained to control a so_follower robotic arm to press 22 musical notes from C3 to A#4. Built on lerobot/smolvla_base (0.45B parameters) and fine-tuned using the LeRobot framework on 755 episodes with three cameras. Ideal for robotics enthusiasts and owners of SO-101 robotic arms who want to run control policies on regular consumer hardware.

Key features

  • Understands text prompts to press 22 specific notes (for example, press D4, press G#3, press A#4).
  • Accepts a 6-value robot state vector and streams from three cameras at 256x256 pixel resolution as input.
  • The compact 0.45B parameter size allows running the model on consumer hardware.
  • Trained on the PawelLach/so101-keys101-22keys dataset (755 episodes, 39,790 frames at 30 FPS).

What you can do

  • Run so_follower robot control via the lerobot-rollout CLI utility to press specified notes based on text prompts.
  • Use the checkpoint or training configuration in LeRobot to experiment with imitation learning on keyboard instruments.

Example

lerobot-rollout \
  --strategy.type=base \
  --robot.type=so_follower \
  --robot.port= \
  --policy.path=PawelLach/smolvla-keys101-22keys \
  --task="press D4" \
  --duration=60
Read moreDetails
Base model: lerobot/smolvla_base (0.45B parameters), Apache-2.0 license.
Training: 6,000 steps, batch size 64, AdamW optimizer, learning rate 0.0001, LeRobot 0.7.1.
Limitations: evaluation results on a real robot have not yet been published by the author.
Discuss Telegram 0

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.

    # PawelLach/smolvla-keys101-22keys
    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 PawelLach/smolvla-keys101-22keys --local-dir ~/models/smolvla-keys101-22keys
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to smolvla-keys101-22keys
  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-keys101-22keys

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

PawelLach/smolvla-keys101-22keys
Author
PawelLach
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments