You can now convert notebooks directly into training jobs to easily run independent training experiments while working on your projects. In contrast to copying the notebook into another notebook job, training jobs will run autonomously, send their output to the location you specify, and automatically terminate when finished.

​proxiML notebooks can now be forked into new instances to enable easy parallel experimentation. Unlike other cloud notebooks, when you fork a proxiML notebook, the entire working directory is copied. All datasets, checkpoints, and other data are copied into the new notebook.

Persistent Datasets just got even better. Not only can you use the same dataset across many jobs in parallel at no additional charge, now you can attach multiple datasets to a single job for free. If that wasn't enough, you can now dynamically change the datasets attached to any notebook job as your needs evolve through the model development process. Additionally, more options have been added for job base environments, allowing you to save time and storage quota by using specific versions of popular frameworks.

Customers using proxiML to compete in Kaggle competitions or using public Kaggle datasets for analysis can now directly populate proxiML datasets from Kaggle competitions or datasets, as well as automatically load their Kaggle account credentials into notebook and training jobs to use for competition or kernel submissions.

Training jobs' worker log output can now be viewed centrally from the proxiML platform in real-time. Keep an eye on all your job workers' training progress at the same time, so you can stop them early if they are no longer making progress.

proxiML customers can now select from a variety of popular public datasets when starting a notebook or training jobs. There is no storage cost for using these datasets, no matter how many jobs you attach them to.

The proxiML platform experience has been redesigned to make it easier to manage notebooks, training jobs, and datasets independently. Additionally, Notebooks are now directly access from the web interface instead of launched through the connection utility.

The proxiML platform now allows customers to store datasets permanently and reuse those datasets for as many notebook and training jobs as desired.

You can now run proxiML serverless deep learning training jobs on model code hosted in private git repositories that support SSH authentication.

proxiML training jobs can now run on data stored in Google Cloud Storage and upload their results to the same or another bucket. GCP access credentials can also be attached to notebooks and training job workers to provide them with easy access to other GCP services.