Skip to content
Other

reachy-mini-gesture-intent-lora-v8-20261117

LoRA for a robotics gesture planner

1 min readupdated

This is a LoRA (a lightweight add-on for a base model) for the binhpham/reachy-mini-motion-planner-4b robotics planner, trained to improve intent recognition and gesture generation. The adapter helps the robot execute joint tasks more accurately while maintaining motion safety. An experimental candidate at step 20 shows a sharp jump in benchmark accuracy compared to the default package. It is a solid fit for robotics researchers and developers working with the Reachy Mini platform.

Key features

  • Major boost to the joint task and safety metric, reaching 90.38%
  • Improved score for natural and explicit joint movements
  • Lower average planner and compiler latency at 1.801 s
  • Evaluated across 200 tasks using bootstrap sampling and trajectory analysis

What you can do

  • Integrate the LoRA with the base planner to accurately reproduce ten familiar robot gestures
  • Use the intent extractor and compiler from the companion experiment dataset
  • Test safe trajectory generation for robotics hardware experiments

Example

Use the components from the experimental dataset to run inference and compile gestures alongside the loaded peft model.

Read moreDetails
Joint task and safety: 90.38% vs. 28.25% for default
Explicit articulation: 99.75% (up by 62.12 percentage points)
Latency: 1.801 s on average for the selected candidate
Library: peft
Discuss Telegram 0 32

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 8.6 GB · ready files9.9 GB
  • FP8 / INT8recommendedweights 4.3 GB5.3 GB
  • 4-bitweights 2.5 GB3.2 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 3080

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 reachy-mini-motion-planner-4b.

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.

    # diwank/reachy-mini-gesture-intent-lora-v8-20261117
    # ===== 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 binhpham/reachy-mini-motion-planner-4b --local-dir ~/models/reachy-mini-motion-planner-4b
    hf download diwank/reachy-mini-gesture-intent-lora-v8-20261117 --local-dir ~/models/reachy-mini-gesture-intent-lora-v8-20261117
    # 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

binhpham/reachy-mini-motion-planner-4b

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

  1. Qwen3.5-4B-Basefamily root
  2. Qwen3.5-4B
  3. reachy-mini-motion-planner-4b
  4. reachy-mini-gesture-intent-lora-v8-20261117

New posts about versions of Qwen3.5-4B-Base (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

diwank/reachy-mini-gesture-intent-lora-v8-20261117
Author
diwank
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

320

Ask AI about this model

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

Comments