Skip to content
Robotics

chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k

a pi0.5 model checkpoint for the robotic cable insertion task

2 min readupdated

This model is a checkpoint (at 40,000 steps) of the pi0.5 robotics policy, fine-tuned on a demonstration dataset for cable insertion in the GRC task using the LeRobot framework. It takes images from three cameras and an 18-dimensional state, predicting intermediate subtasks with a built-in VLM (Vision-Language Model) and outputting an action sequence for the manipulators. The model is tailored for researchers and engineers working on robot control using the LeRobot library.

Key features

  • Base model: Based on lerobot/pi05_base using the pi05_subtask plugin.
  • Inputs: Uses data from three cameras (two d405 and one zed_mini) and an 18-dimensional state vector.
  • Subtask prediction: The VLM records the current subtask, after which the action expert builds the trajectory.
  • Training parameters: Batch size 128 (64 per GPU x 2 B200), AdamW optimizer, bf16 precision.

What you can do

  • Run policy: Load the checkpoint via the PI05SubtaskPolicy class to control the robotic setup.
  • Hardware integration: Pass camera data and joint states to perform the cable insertion task.
  • Test subtasks: Use subtask prediction to automate multi-stage manipulation tasks with manipulators.

Example

import torch, lerobot_policy_pi05_subtask
from lerobot.policies.factory import make_pre_post_processors
from lerobot_policy_pi05_subtask import PI05SubtaskPolicy
policy = PI05SubtaskPolicy.from_pretrained("chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k")
pre, post = make_pre_post_processors(policy.config, pretrained_path="chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k")
pad = torch.zeros(18, dtype=torch.bool)
batch = pre({"observation.state": state18, "observation.images.d405_left": img_left, "observation.images.d405_right": img_right, "observation.images.zed_mini": img_zed, "task": "insert the cable into the control module", "state_dim_is_pad": pad})
actions = post(policy.predict_action_chunk(batch))
Read moreDetails
Framework: LeRobot 0.6.2 pi05_subtask
Training: 40,000 steps, train loss 0.002
Hardware: 2 x B200 GPU (DDP)
Action space: 18-dimensional, chunk 30 steps
Discuss Telegram 0 19

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 15 GB · ready files17 GB
  • BF16weights 7.7 GB9.0 GB
  • FP8 / INT8recommendedweights 3.9 GB4.7 GB
  • 4-bitweights 2.2 GB2.9 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 · F32 — from RTX 4090

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.

    # chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k
    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 chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k --local-dir ~/models/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k
    # 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

  1. paligemma-3b-pt-224family root
  2. pi05_base
  3. task2_full_grc_subtask_rel_chunk30_bs128_1120_40k

New posts about versions of paligemma-3b-pt-224 (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 12All on Hugging Face
LoRAs and adapters 12All on Hugging Face

About the model

chomeed/task2_full_grc_subtask_rel_chunk30_bs128_1120_40k
Author
chomeed
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