Frequently Asked Questions

What is WeightsLab, and who is it for?

WeightsLab is an interactive AI model development platform for efficient training, fine-tuning and de-bugging of AI models.
It’s built for AI researchers, engineers, and developers who want to work faster and smarter, with flexibility and full control over data and model architectures.

How do I access WeightsLab?

Sign up for our waitlist. If you're a university or organization looking for wide scale access, contact hello@graybx.com.

What do I need to start training?

A dataset of supervised learning examples. After selecting and/or defining a base model to train on, WeightsLab provides simple and intuitive graphics and user interfaces to interact with data samples and model weights, in-training. Check our Github for examples to get started.

Do I need to manage my own training/GPU infrastructure?

Yes. WeightsLab is built for on-premise use, so everything from storing data and access to GPUs will be in your control.
You can run training on your own hardware — from a single GPU to a growing cluster.

What model architectures does WeightsLab support?

We currently support computer vision models (non-transformers), with more architectures coming soon as we expand the platform.

Who owns the data and models, and is my data safe?

You retain 100% ownership of your data, models, and weights. All training happens locally — no data leaves your local environment, ensuring full control over the data and privacy.

How much do I have to pay to use WeightsLab?

WeightsLab will be free to start. To learn more about our future pricing plans, contact us at hello@graybx.com