### Start DROID Inference Server

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Starts the DROID inference server with default settings on all interfaces.

```bash
uv run python examples/droid/host_server_droid.py --host 0.0.0.0
```

--------------------------------

### Start YAM Inference Server

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Starts the YAM inference server with default settings on all interfaces.

```bash
uv run python examples/yam/host_server_yam.py --host 0.0.0.0
```

--------------------------------

### Install uv package manager

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Installs the uv package manager using a curl script and reloads the shell environment. Verifies the installation by printing the uv version.

```bash
curl -LsSf https://astral.sh/uv/install.sh | sh
exec $SHELL          # reload PATH so the `uv` binary is picked up
uv --version
```

--------------------------------

### Create and sync project environment

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Creates a virtual environment named .venv/ and installs all project dependencies as defined in pyproject.toml using uv. It then verifies CUDA availability and the GPU name.

```bash
uv sync                  # creates .venv/ and installs all deps
uv run python -c "import torch; print(torch.cuda.is_available(), torch.cuda.get_device_name(0))"
# expected: True NVIDIA RTX A6000
```

--------------------------------

### Start DROID inference server

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Starts the DROID inference server using uv run, specifying the host, port, and data type. The --dtype flag can be adjusted based on available VRAM.

```bash
# DROID (Franka)
uv run python examples/droid/host_server_droid.py --host 0.0.0.0 --port 8000 --dtype bfloat16
```

--------------------------------

### Start DROID Server with Full Precision and CUDA Graphs

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Starts the DROID inference server using float32 precision and enabling CUDA graphs for potential performance gains.

```bash
uv run python examples/droid/host_server_droid.py --dtype float32 --cuda-graph
```

--------------------------------

### Install/Refresh Dependencies

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Installs or refreshes project dependencies using uv.

```bash
uv sync
```

--------------------------------

### Start Bimanual YAM inference server

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Starts the Bimanual YAM inference server using uv run, specifying the host, port, and data type. The --dtype flag can be adjusted based on available VRAM.

```bash
# Bimanual YAM
uv run python examples/yam/host_server_yam.py --host 0.0.0.0 --port 8202 --dtype bfloat16
```

--------------------------------

### Initialize LeRobot Submodule

Source: https://github.com/allenai/molmoact2/blob/main/README.md

After cloning the repository, initialize the LeRobot submodule recursively. This command ensures all necessary LeRobot components are downloaded and set up.

```bash
git submodule update --init --recursive
cd lerobot
```

--------------------------------

### Download YAM Model

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Pre-caches the YAM model weights from Hugging Face.

```bash
uv run hf download allenai/MolmoAct2-BimanualYAM
```

--------------------------------

### Download DROID Model

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Pre-caches the DROID model weights from Hugging Face.

```bash
uv run hf download allenai/MolmoAct2-DROID
```

--------------------------------

### Download MolmoAct2 checkpoints

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Downloads the MolmoAct2 checkpoints for both DROID and Bimanual YAM configurations using the hf download command. Ensures fast parallel downloads by setting HF_HUB_ENABLE_HF_TRANSFER.

```bash
export HF_HUB_ENABLE_HF_TRANSFER=1                       # fast parallel download
uv run hf download allenai/MolmoAct2-DROID               # for the DROID server
uv run hf download allenai/MolmoAct2-BimanualYAM         # for the YAM server
```

--------------------------------

### Allow LAN Connections to MolmoAct2 Ports

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Configure the Uncomplicated Firewall (ufw) to allow incoming TCP connections on specific ports from a given subnet. This is necessary if clients on the local area network cannot connect to the MolmoAct2 service.

```bash
sudo ufw allow from <subnet> to any port 8000 proto tcp   # DROID
sudo ufw allow from <subnet> to any port 8202 proto tcp   # YAM
```

--------------------------------

### Check DROID Server Health

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Sends a request to the DROID server's health endpoint.

```bash
curl http://<host>:8000/act
```

--------------------------------

### Check YAM Server Health

Source: https://github.com/allenai/molmoact2/blob/main/CLAUDE.md

Sends a request to the YAM server's health endpoint.

```bash
curl http://<host>:8202/act
```

--------------------------------

### MolmoAct2 Citation Information

Source: https://github.com/allenai/molmoact2/blob/main/README.md

BibTeX entry for citing the MolmoAct2 model in academic publications. Includes title, authors, year, arXiv ID, and URL.

```bibtex
@misc{fang2026molmoact2actionreasoningmodels,
      title={MolmoAct2: Action Reasoning Models for Real-world Deployment},
      author={Haoquan Fang and Jiafei Duan and Donovan Clay and Sam Wang and Shuo Liu and Weikai Huang and Xiang Fan and Wei-Chuan Tsai and Shirui Chen and Yi Ru Wang and Shanli Xing and Jaemin Cho and Jae Sung Park and Ainaz Eftekhar and Peter Sushko and Karen Farley and Angad Wadhwa and Cole Harrison and Winson Han and Ying-Chun Lee and Eli VanderBilt and Rose Hendrix and Suveen Ellawela and Lucas Ngoo and Joyce Chai and Zhongzheng Ren and Ali Farhadi and Dieter Fox and Ranjay Krishna},
      year={2026},
      eprint={2605.02881},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2605.02881},
}
```

--------------------------------

### Health check for Bimanual YAM server

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Performs a health check on the Bimanual YAM inference server by sending a curl request to the /act endpoint. Displays the server status and configuration details.

```bash
curl http://<lan-ip>:8202/act
# YAM:   {"status":"ok","repo_id":"allenai/MolmoAct2-BimanualYAM","norm_tag":"yam_dual_molmoact2","num_cameras":3,"state_dim":14,...}
```

--------------------------------

### Health check for DROID server

Source: https://github.com/allenai/molmoact2/blob/main/README.md

Performs a health check on the DROID inference server by sending a curl request to the /act endpoint. Displays the server status and configuration details.

```bash
curl http://<lan-ip>:8000/act
# DROID: {"status":"ok","repo_id":"allenai/MolmoAct2-DROID","norm_tag":"franka_droid",...}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.