Keepsake 0.4.2

Fix a bug where Keepsake would check out even if you told it not to. #567.

Run pip install -U keepsake to upgrade.

Keepsake 0.4.1

Just a few bugs squished:

• Prefix some private Python APIs with underscores so they don't mess with IDE auto-complete. #542 (thanks @gabrielmbmb!)
• Fix different plots being written on the same plot with incorrect axes. #553 (thanks @enochkan!)
• Use AWS_DEFAULT_REGION environment variable to determine what region to create S3 buckets in. #561
• Correctly throw error if path passed to checkpoint() does not exist. #556
• Fix an issue where if only a single file was saved, then checkout would fail. #560
• Bump dependencies. (thanks @dependabot!)

Run pip install -U keepsake to get the new goodness.

Keepsake 0.4.0

With this release, we've renamed Replicate to Keepsake.

We're going to use "Replicate" as the name for our broader project to make machine learning reproducible. We think that tools should do one thing well, and we want to create space for multiple tools.

Migration is simple:

• Use pip install keepsake to install or use keepsake in requirements.txt
• Rename replicate to keepsake in your code, both where it is imported and where it is used
• If you use the Keras or PyTorch Lightning callbacks, they're now both called KeepsakeCallback
• Rename replicate.yaml to keepsake.yaml
• Rename .replicateignore to .keepsakeignore

If you just want to stick with the replicate package that's no problem, but you'll just miss out on new stuff in the future.

No new features in this release. Keep on trackin'.

0.3.3

New features:

• Limit the number of Python packages that are shown by default in replicate show to make it easier to read

Bug fixes:

• Rename replicate ls filter name started to created for consistency -- thanks to @murthy95 for this contribution!
• Fix Python version not being saved in the experiment
• Display output immediately when using the CLI with S3 or GCS to make it more responsive

0.3.2

Fix a bug where Replicate would consume lots of memory when uploading large files

0.3.1

Fix a missing Python dependency.

Replicate 0.3.0

This is the first big release since we launched Replicate, and we've got a smorgasbord of new features for you.

In case you're landing here for the first time, Replicate is a version control system for machine learning models.

Some highlights:

Files are uploaded in the background

When you pass files to replicate.init() or experiment.checkpoint(), these will now be uploaded in the background while your training continues. Using Replicate won't make your training any slower.

This also involved a lot of behind-the-scenes improvements which have made things faster and more robust. For example, opening files from Python is now much faster.

Easier to read experiment list

The output of replicate ls has been redesigned so it works better with more experiments and params:

Check out individual files

You can now check out individual files instead of everything inside a checkpoint and its experiment. For example, you can run this to just check out the file model.pth:

replicate checkout 3b453f0 --path model.pth

Thanks to @VastoLorde95 for working on this!

PyTorch Lightning integration

Replicate now includes a callback that makes it much easier to use with PyTorch Lightning. Full details are in the documentation.

Thanks @gan3sh500 for working on this!

And a few more things

Some more features:

• experiment.checkpoint() automatically generates the step number if you don't specify it. (Thanks @justinchuby!)
• The Python version is now recorded when you run your experiment. One more step towards making your training environment reproducible. (Thanks @kvthr!)

And a few bugs squashed:

• Checkpoints are no longer counted as best if the primary metrics is NaN or None. (Thanks @ryanbloom!)
• Fix ID generation when random seed is fixed.

Thanks to everyone who suggested features, complained about broken things, and help build things!

Run pip install -U replicate to upgrade. Or, add replicate~=0.3.0 to requirements.txt if you're using that.

• If you haven't tried Replicate yet, check out the documentation to get started.
• Join us in Discord if you have any questions, want to get involved, or just want to say hi!
• Follow us on Twitter to keep up to date