Welcome to the 4th edition of the Unweave Changelog. It’s the age of AI. The progress in AI art in the last few months has been 🤯. Since we have the perfect platform with our Changelog, we’ll be featuring AI generated art as the cover from now on. If you’d like to train your own AI art generator reach out and we’ll get you setup with Unweave ASAP!
Docs website is out
We’re debuting into the publishing business with our brand new docs website. We’re starting out with a starter template to help you train an MNIST classifier from scratch within 2 minutes (literally). We’ll keep adding more templates each week. If you have a request for a starter you’d like to see, hit us up and we’ll make it happen.
Versioned Zepl Output
The feedback we often get from users is how difficult it is to version the models they train. As of today, Unweave does this for you out of the box! When you run your code with Unweave, we automatically snapshot the output you save to the
uwstore/output directory. In the image below, for example, the zepl on the left was aborted by the user and therefore had no output. The zepl on the right, ran to completion and saved a checkpoint file
minst_ccn.pt to the
uwstore/output directory. This data is automatically snapshotted by Unweave. You can retrieve it for every zepl you ever ran from the Dashboard or the CLI.
store Command in the CLI
Last week, we announced the
uwstore: an S3 bucket mounted at the base of your repository. Today, we’re adding the
unweave store command to the CLI to make it even easier to manage your data. You can now upload and download files to the store using the
unweave store download and
unweave store upload commands. You can also list the contents of your store with
uwstore store list. Give it a whirl and let us know what you think!
Usage: unweave store [command] Available Commands: download Download an object from the uwstore list List objects in the uwstore upload Upload a local directory or file to the uwstore Flags: -h, --help help for store Global Flags: -p, --path string Path to an Unweave project to run -t, --token string Use a specific token to authenticate - overrides login token Use "unweave store [command] --help" for more information about a command.
We’re still in early access. If you’d like to be one of the first to try Unweave and help us shape the product, get in touch and we’ll get you setup asap ❤️