SageMaker JumpStart model zoo

There are more than 200 popular prebuilt and pretrained models in SageMaker JumpStart for you to use out of the box or continue to train for your use case. What are they good for? Training an accurate deep learning model is time consuming and complex, even with the most powerful GPU machine. It also requires large amounts of training and labeled data. Now, with these models that have been developed by the community, pretrained on large datasets, you do not have to reinvent the wheel.

Model collection

There are two groups of models: text models and vision models in SageMaker JumpStart model zoo. These models are the most popular ones among the ML community. You can quickly browse the models in SageMaker JumpStart and select the one that meets your needs. On each model page, you will see an introduction to the model, its usage, and how to prepare a dataset for fine-tuning purposes. You can deploy models into AWS as a hosted endpoint for your use case...