Skip to content
/ jako Public

Parallelization of Hyperparameter Experiments with Talos

License

Notifications You must be signed in to change notification settings

autonomio/jako

Repository files navigation


Jako

Parallelization of Hyperparameter Experiments with Talos

JakoKey FeaturesExamplesInstallSupportDocsIssuesLicenseDownload


Jako makes it straightforward to distribute Talos experiments across one or more remote machines without asking you to change anything in the way you are already working with Talos.

Jako

Jako solves the problem of wanting to run a single Talos experiment on one or more remote machines:

  • Easily start distributed Talos experiments
  • Access experiment results in real-time through centralized datastore
  • Takes minutes to implement
  • No new syntax or anything else to learn
  • Adds no overhead to your current Talos workflow

Jako is built to work exclusively with Talos.


🔧 Key Features

  • jako.DistributedScan() works exactly like talos.Scan()
  • Distribute experiments across one or more remote machines
  • Local machine can be included in the experiment
  • Zero-configuration; add new machines simply by adding them to a config file
  • Manage keys, passwords and other details in a single config file or a dictionary
  • Optionally run DistributedScan() through talos.DistributedScan()

Jako works on Linux, Mac OSX, and Windows systems and can be operated cpu, gpu, and multi-gpu systems.


▶️ Examples

All of the below Talos examples will work, simply change talos.Scan() to talos.DistributedScan() or jako.DistributedScan() and you're good to go.

Get the below code here. More examples further below.

The Simple example below is more than enough for starting to use Talos with any Keras model. Field Report has +2,600 claps on Medium because it's more entertaining.

Simple [1-2 mins]

Concise [~5 mins]

Comprehensive [~10 mins]

Field Report [~15 mins]

For more information on how Talos can help with your Keras, TensorFlow (tf.keras) and PyTorch workflow, visit the User Manual.

You may also want to check out a visualization of the Talos Hyperparameter Tuning workflow.


💾 Install

Stable version:

pip install jako

Daily development version:

pip install git+https://github.com/autonomio/jako


💬 How to get Support

I want to... Go to...
...troubleshoot Docs · GitHub Issue Tracker
...report a bug GitHub Issue Tracker
...suggest a new feature GitHub Issue Tracker
...get support Stack Overflow

📢 Citations

If you use Talos for published work, please cite:

Autonomio Talos with Jako [Computer software]. (2022). Retrieved from http://github.com/autonomio/jako.


📃 License

MIT License

About

Parallelization of Hyperparameter Experiments with Talos

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages