Instructions to use facebook/sam3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use facebook/sam3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("mask-generation", model="facebook/sam3")# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModel processor = AutoProcessor.from_pretrained("facebook/sam3") model = AutoModel.from_pretrained("facebook/sam3", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Request for manual review / reset of my access request for facebook/sam3
Hello SAM 3 Team,
My access request for facebook/sam3 was rejected. I would like to kindly ask for a manual review, or a reset of my request, so that I can resubmit it with complete and accurate information. I apologize if my original submission was incomplete or unclear.
Intended use
We are running RLE-Bench (arXiv:2609.34210), an open-source benchmark that evaluates whether AI coding agents can solve robot learning engineering tasks in simulation. In its kitchen manipulation tasks, built on the RoboCasa simulator, SAM 3 is one of the perception tools available to the agent. It performs text-prompted segmentation on simulator-rendered images, alongside Contact-GraspNet for grasp proposals.
How the model will be used
- Inference only, with frozen weights at a fixed revision. No fine-tuning or retraining.
- Inputs are synthetic images rendered by the simulator. No images of people and no personal data.
- Used solely for research and benchmark evaluation. [It is not part of any product or service.]
- The weights will stay within our access-controlled internal evaluation environment and are loaded offline. They will not be published, shared, or redistributed outside our team.
- I have read the SAM License and will comply with it, including its trade control provisions, as well as the Hugging Face Terms of Service.
I would be happy to provide any additional information you may need. Thank you very much for your time and consideration.
Best regards,
Hugging Face username: shuyunljf