OpenDM

English | 简体中文

Introduction

DM0.5 is Dexmal's next-generation Vision-Language-Action model (VLA) for open-world robot control. It builds on the native embodied modeling approach introduced by DM0, with systematic upgrades for open-ended instructions, long-horizon tasks, dynamic disturbances, and multi-embodiment robot control.

OpenDM provides DM0.5 model weights, training and inference scripts, dataset registration examples, and evaluation workflows for researchers and developers to train, fine-tune, evaluate, and deploy the model.

News

• [2026-07-09] DM0.5 is officially released. Read the technical blog for more details.

Models

Model	Description	Checkpoint
DM05	Base DM0.5 model for fine-tuning	🤗 Dexmal/DM05
DM05-libero	LIBERO fine-tuned DM0.5 model for evaluation	🤗 Dexmal/DM05-libero

Example checkpoint download:

huggingface-cli download Dexmal/DM05 --local-dir ./checkpoints/DM05

Quick Start

We recommend using Docker to set up the runtime environment first, which helps avoid version mismatches across CUDA, PyTorch, flash-attn, and other dependencies on the host machine.

Requirements

System requirements:
Ubuntu 20.04 / 22.04
NVIDIA GPU
NVIDIA Driver
Docker
NVIDIA Container Toolkit
Conda (optional, only required for local pip installation)

Recommended GPUs:
RTX 4090, A100, H100, H20
8 GPUs are recommended for training, and 1 GPU is sufficient for deployment inference.

Docker Installation

git clone https://github.com/dexmal/opendm.git
cd opendm

docker run -it --rm --gpus all --network host \
  --name opendm \
  --shm-size=16g \
  -v "$PWD":/app/opendm \
  -w /app/opendm \
  dexmal/opendm:latest /bin/bash

# Run from the OpenDM repository root inside the container.
conda activate opendm
pip install -e .

Local Installation

conda create -n opendm python=3.10 -y
conda activate opendm

pip install torch torchvision \
  --index-url https://download.pytorch.org/whl/cu128

pip install ninja packaging
MAX_JOBS=2 pip install flash-attn --no-build-isolation

# Enter the OpenDM repository root.
cd opendm
pip install -e .

Inference

After installing the environment and initializing the source code, you can start the model inference service. The service loads the specified checkpoint and exposes an HTTP endpoint for benchmark clients or other applications to request action predictions. Use a checkpoint that contains norm_stats.json, or make sure the matching stats already exist under ./norm_stats/.

script/dm05_launcher.sh \
  --task inference \
  --nproc_per_node 1 \
  --model-config.model-name-or-path ./checkpoints/DM05 \
  --model-config.chunk-size 50 \
  --inference-config.port 7891

Arguments:

• --task: task type. Use inference for inference.
• --nproc_per_node: number of GPUs on a single node. 1 GPU is sufficient for inference.
• --model-config.model-name-or-path: model checkpoint path.
• --model-config.chunk-size: action chunk length.
• --inference-config.port: inference service port.

During inference, the service first looks for norm_stats.json in the checkpoint directory. If it is not found, it falls back to the matching file under ./norm_stats/, which is normally generated during training for the same dataset, action mode, and chunk size.

After the service starts, send a test request to verify that the endpoint returns a valid response:

bash tests/curl_demo.sh http://SERVER_IP:7891/process_frame

/process_frame accepts a multipart/form-data request:

• text: task instruction.
• states: JSON array of the current robot state. The dimension and order must match the model's training and normalization statistics.
• image: image files, one field per configured image key. The order must match --inference-config.image-keys.
• robot_type: optional built-in robot type. Currently only DOS W1 is supported. It provides the robot state description when relative actions need to be converted back to absolute actions.
• control_mode and speed: text conditioning fields required when directly serving the pretrained Dexmal/DM05 model. They are normally not required for SFT checkpoints unless your SFT data was trained with the same fields.

A successful response has the following shape.

{
  "response": [
    [0.012, -0.034, 0.18, "..."],
    [0.015, -0.031, 0.17, "..."],
    ...
  ]
}

Training

Data Preparation

Prepare data files and dataset configuration according to the dexbotic Data Guide. Make sure --data-config.dataset-name in the training command matches the registered dataset name.

The training script selects a dataset through --data-config.dataset-name. Before training, register your dataset in the project dataset registry. We recommend using an existing file such as opendm/dataset/demo.py as a reference, then creating a new dataset config file such as opendm/dataset/my_robot.py and updating the dataset name, data paths, image keys, and state description.

# opendm/dataset/my_robot.py

from opendm.constants.robot import RobotStateDesc
from opendm.dataset.register import register_dataset

MY_ROBOT_STATE_DESC = (
    [RobotStateDesc.JOINT] * 6
    + [RobotStateDesc.GRIPPER]
    + [RobotStateDesc.JOINT] * 6
    + [RobotStateDesc.GRIPPER]
)

register_dataset(
    {
        "my_robot": {
            "jsonl_dir": "./assets/my_robot/",
            "image_dir": "./assets/my_robot/",
            "image_keys": ["images_1", "images_2", "images_3"],
            "state_desc": MY_ROBOT_STATE_DESC,
        },
    }
)

Field descriptions:

• my_robot: dataset name registered in the dataset registry. Use it with --data-config.dataset-name my_robot.
• jsonl_dir: directory containing training jsonl files.
• image_dir: directory containing image files.
• image_keys: image field names to load from the dataset.
• state_desc: semantic description of each state/action dimension, such as robot joints and grippers.

During training, if the corresponding normalization statistics file does not exist, the script automatically computes it from the current dataset, action mode, and chunk size, then saves it under ./norm_stats/.

Start Training

After environment setup, source initialization, and data preparation, start model training. The training script reads the specified dataset configuration, loads the base checkpoint, and starts training according to the configuration.

script/dm05_launcher.sh \
  --task train \
  --nproc_per_node 8 \
  --data-config.dataset-name my_robot \
  --model-config.model-name-or-path ./checkpoints/DM05 \
  --model-config.chunk-size 50

Arguments:

• --task train: run in training mode.
• --nproc_per_node 8: number of training processes on a single node, usually matching the number of GPUs.
• --data-config.dataset-name my_robot: dataset name for training. It must match the project dataset configuration.
• --model-config.model-name-or-path ./checkpoints/DM05: initial model checkpoint path.
• --model-config.chunk-size 50: action chunk length predicted by the model.

Training logs will include data loading, model initialization, loss values, and checkpoint saving. Before running a full training job, verify that the data path, model checkpoint path, and GPU count are correctly configured.

DM05 SFT with Demo and Custom Data

Start by running a complete DM05 SFT workflow with the built-in demo data and playground/dm05_sft_demo.py. After you are familiar with the data format, normalization statistics, training, inference, and service validation flow, replace the demo dataset with your own robot data for SFT. See DM05 SFT and Validation Guide.

LIBERO Fine-Tuning Reference

Use the DM05 LIBERO Training and Evaluation Guide as an end-to-end reference for fine-tuning DM05. It covers data and model preparation, SFT training, inference service startup, and benchmark evaluation, and can help you adapt DM05 to your own robot datasets.

Guides

• Download models: see Models or visit Dexmal Hugging Face.
• Prepare data: see the Data Guide.
• Start inference service: see Inference.
• DM05 SFT with demo or custom data: see DM05 SFT and Validation Guide.
• LIBERO training and evaluation: see the DM05 LIBERO Training and Evaluation Guide; for LoRA SFT, see DM05 LIBERO LoRA Training.

Community and Support

• Learn more about Dexmal products and model updates on the Dexmal website.
• Get DM model weights from Dexmal Hugging Face.
• If you encounter issues, please report them through GitHub Issues.
• For further discussion, scan the WeChat QR code to contact us.

We will continue to release more model weights, technical documentation, and examples. If this project is helpful to you, please consider giving us a star on GitHub . Your support helps us move forward.

License

This project is licensed under the Apache-2.0.