Skip to content
Robotics

nberezny/smolvla_smoke

Robotics policy for uFactory Lite 6

2 min readupdated

This model is a specialized control policy for robots, trained using the LeRobot library (a machine learning toolkit for real-world robotics). Built on the compact SmolVLA vision-language-action architecture, it is designed for a specific manipulation task: sequentially stacking blocks of different colors. The policy takes mechanism states and images from three cameras as input and outputs control actions for the robot. This release will be useful for researchers and engineers working on imitation learning for robotics in simulation.

Key features

  • Built on the efficient SmolVLA architecture (a compact vision-language-action model)
  • Trained using the LeRobot framework for imitation learning tasks
  • Processes visual data from three different camera angles (overhead, wrist, and state)
  • Configured to work with the uFactory Lite 6 robot simulator

What you can do

  • Run the ready-to-use control policy on a compatible robot or in simulation using LeRobot utilities
  • Use the uaberezny/lite6_stack_dense dataset to study block stacking trajectories
  • Fine-tune the model on your own demonstrations to solve similar manipulation tasks

Example

lerobot-rollout \
  --strategy.type=base \
  --robot.type=ufactory_lite6_sim_hybrid_dense_rel \
  --robot.port= \
  --robot.cameras="{ : {type: opencv, index_or_path: , width: 640, height: 480, fps: 30}, : {type: opencv, index_or_path: , width: 640, height: 480, fps: 30}}" \
  --policy.path=nberezny/smolvla_smoke \
  --task="stack the blocks: red at the bottom, yellow, blue on top" \
  --duration=60
Read moreDetails
License: apache-2.0
Base model: lerobot/smolvla_base
Dataset: 4,924 episodes, 143,843 frames at 20 FPS
Training parameters: 200 steps, batch size 32, adamw optimizer, learning rate 0.0001
Discuss Telegram 0 19

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.

    # nberezny/smolvla_smoke
    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 nberezny/smolvla_smoke --local-dir ~/models/smolvla_smoke
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to smolvla_smoke
  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_smoke

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

nberezny/smolvla_smoke
Author
nberezny
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

190

Ask AI about this model

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

Comments