To assist developers in learning about and experimenting with machine learning, AWS has launched SageMaker Studio Lab, a free offering. Users may train models on CPUs and GPUs, use 15 GB of persistent storage, and utilise the JupyterLab IDE, all of which are included with SageMaker Studio Lab.
To construct data analytics, scientific computing, and machine learning projects with notebooks that can be quickly imported and exported through the Git repo or a private Amazon S3 bucket, SageMaker Studio Lab contains all the necessary components.
With free CPU/GPU access, SageMaker Studio Lab serves as a substitute for the well-known Google Colab environment.
A visual, no-code tool called SageMaker Canvas is another improvement for AWS SageMaker. By merging datasets, combining data sources in the cloud or on-premises, and training models once fresh data is available, Canvas enables business analysts to create machine-learning models and make predictions. The new service has a wizard-style user interface for data submission, model training, and prediction tasks.
AWS also unveiled a new turnkey solution that makes use of a skilled staff to supply high-quality training datasets, freeing businesses from the burden of maintaining their own labelling systems. The SageMaker Ground Truth Plus is the name of the new service. Data scientists have the ability to collaborate with labelers both inside and outside of their business using SageMaker Ground Truth.
Another new SageMaker feature, SageMaker Training Compiler, attempts to speed up deep-learning model training by automatically compiling developers’ Python source code and creating GPU kernels tailored for their model. The SageMaker machine-learning GPU instances are used more effectively by the compiler to optimise deep-learning models to speed up training. The SageMaker platform offers the service without charge.
Finally, there is the new SageMaker Serverless Inference option, which enables customers to deploy machine learning models for inference without having to set up or maintain the supporting infrastructure. SageMaker automatically configures, adjusts, and deactivates computing capacity using Serverless Inference based on the number of inference requests. Customers simply pay for the time it takes the inference code to execute and the volume of data it processes; they do not pay for downtime.
A free SageMaker Studio Lab account can be requested. To guarantee a high standard of experience for clients, the number of new account registrations will be restricted. Sample notebooks are available in the Studio Lab GitHub repository.