Skip to content
Robotics

VLA_Samer_08.11.2026_400eps_policy_XVLA

a VLA policy for the UR5 robot

2 min readupdated

This model is a VLA policy (Vision-Language-Action — an AI model that links vision, text instructions, and physical actions) built to control a ur5 industrial robotic arm. Trained on the 0.88B-parameter lerobot/xvla-base model using the LeRobot framework, it handles a specific pick-and-place task: using a vacuum gripper to pick up a CPU and place it into a box. The X-VLA architecture uses flow-matching alongside learnable Soft Prompts (trainable prompt vectors) that adapt a single unified model to a specific robot's sensor setup and action space. It is aimed at robotics engineers and researchers working with Universal Robots manipulators.

Key features

  • Built on lerobot/xvla-base (0.88B parameters) using LeRobot version 0.6.2.
  • Takes feeds from three cameras (two 256×256 streams and one 224×224 stream) plus an 8-dimensional robot state vector.
  • Outputs a 7-dimensional action vector to control picking and placing operations.
  • Trained on a 400-episode dataset (259,635 frames recorded at 30 FPS).

What you can do

  • Deploy the pre-trained policy on a real UR5 arm equipped with a vacuum gripper to sort and pack CPUs.
  • Use the weights and configuration as a reference baseline when training custom X-VLA policies in LeRobot.

Example

lerobot-rollout \
  --strategy.type=base \
  --robot.type=ur5 \
  --robot.port= \
  --policy.path=SamerAyyad/VLA_Samer_08.11.2026_400eps_policy_XVLA \
  --task="Pick up the CPU with the suction gripper and place it in the box" \
  --duration=60
Read moreDetails
Training: 20,000 steps, batch size 8, xvla-adamw optimizer, learning rate 0.0001.
Real-world robot evaluation results are not yet listed on the model card.
License: Apache-2.0.
Discuss Telegram 0 26

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.6 GB · ready files2.3 GB
  • FP8 / INT8recommendedweights 0.8 GB1.4 GB
  • 4-bitweights 0.5 GB1.0 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.

    # SamerAyyad/VLA_Samer_08.11.2026_400eps_policy_XVLA
    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 SamerAyyad/VLA_Samer_08.11.2026_400eps_policy_XVLA --local-dir ~/models/VLA_Samer_08.11.2026_400eps_policy_XVLA
    # 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

New posts about versions of xvla-base (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 1All on Hugging Face
Fine-tunes 12All on Hugging Face

About the model

SamerAyyad/VLA_Samer_08.11.2026_400eps_policy_XVLA
Author
SamerAyyad
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

260

Ask AI about this model

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

Comments