Skip to content
Robotics

longwam-yam-conveyor

Fine-tuned model for a robotic arm on a moving belt

1 min readupdated

This is a specialized checkpoint trained on top of Long-WAM-YAM to control the dual-arm YAM robot in the Isaac Lab simulator. The model tackles grabbing a red cube from a moving conveyor belt traveling at 4–7 cm/s and placing it into a blue container. The repository provides a trained action expert of 1.02B parameters, while the video expert remains frozen. This approach significantly lowers resource requirements when adapting the control policy to specific physical tasks.

Key features

  • Trained on 200 demonstrations with belt speed and lighting randomization
  • Lightweight delta weights (around 2 GB in bf16 format)
  • Multimodal input from top-down cameras and robot wrist cameras
  • Efficient training on two RTX PRO 6000 GPUs in 3 hours and 20 minutes

What you can do

  • Merge the delta weights with the base Long-WAM-YAM model using the merge script
  • Test the control policy in the YAMLab simulator
  • Integrate the model into robotics projects based on Isaac Sim

Example

hf download Efficient-Large-Model/Long-WAM-YAM --revision e104e72dab1991a166e81e2941f7539e4edb8d68 --local-dir base
hf download kabilanKB/longwam-yam-conveyor checkpoints/step_000250/action_step_000250.pt --local-dir .
python eval/thor/merge_action_delta.py base checkpoints/step_000250/action_step_000250.pt yam_conveyor_ft_250
Read moreDetails
Base model: Efficient-Large-Model/Long-WAM-YAM
Parameters: 1.02 B (trainable expert)
Hardware: 2x RTX PRO 6000 (96 GB)
Training duration: 1,216 steps
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.9 GB · ready files2.6 GB
  • FP8 / INT8recommendedweights 1.0 GB1.5 GB
  • 4-bitweights 0.5 GB1.1 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.

    # kabilanKB/longwam-yam-conveyor
    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 kabilanKB/longwam-yam-conveyor --local-dir ~/models/longwam-yam-conveyor
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to longwam-yam-conveyor
  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 Long-WAM-YAM (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 12All on Hugging Face

About the model

kabilanKB/longwam-yam-conveyor
Author
kabilanKB
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