Instructions to use Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps

SGLang

How to use Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps with Docker Model Runner:
```
docker model run hf.co/Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps
```

MolmoAct2 NEXTAGE Forceps

This is a full fine-tune of allenai/MolmoAct2 for a bimanual NEXTAGE robot performing a surgical forceps handover task.

Training

Dataset: Michi-Tsubaki/hand_over_the_forceps_to_the_hand
Training steps: 3,000
Global batch size: 64
Observation cameras: observation.images.top, observation.images.right_wrist
Robot state: observation.state
Control mode: absolute joint position
Action horizon / executed steps: 30 / 30
Normalization tag: nextage_forceps

The repository contains the continuous action expert and the embodiment-specific normalization metadata in norm_stats.json. Loading the model requires trust_remote_code=True because the MolmoAct2 Transformers implementation is included with the checkpoint.

Intended use and safety

This checkpoint is intended for research on the NEXTAGE setup and action convention described above. It is not a general-purpose robot policy. Validate outputs offline and in simulation before hardware use. Apply controller-side joint, velocity, workspace, torque, and contact-force limits; keep an emergency stop available; and operate under human supervision.

References

Downloads last month: -

Safetensors

Model size

5B params

Tensor type

F32

Model tree for Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps

Base model

allenai/MolmoAct2

Finetuned

(42)

this model

Paper for Michi-Tsubaki/MolmoAct2-NEXTAGE-Forceps

MolmoAct2: Action Reasoning Models for Real-world Deployment

Paper • 2605.02881 • Published May 4 • 356