2

SpotML - Cut ML Training Costs by up to 70%

 2 years ago
source link: https://www.spotml.io/
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client
Cut ML Training Costs by up to 70%

Cut ML training costs by up to 70%

SpotML is a command-line tool that lets you train machine learning models on cheap AWS Spot Instances or GCP Preemptible VMs, easily & automatically.

Train your model entirely on Spot Instances, or mix with On-Demand resources to meet your deadlines.
BeforeAfter

How SpotML works

1. Add config file to the root directory

6157f7d18eb7c58fd6e6bc39_Screen%20Shot%202021-10-02%20at%201.08.36%20AM.jpg

2. Run command to sync folder and train

61589a359d44678408d61e38_Screen%20Shot%202021-10-02%20at%2012.43.05%20PM.jpg

3. Run command "spot manage run" to automatically start a remote server. SpotML will wait for the training to complete to stop the server.

Why SpotML?

We make Spot Instances and Preemptible VMs suitable for ML training.

Spot Instances are 70% cheaper than On-Demand instances but are prone to interruptions. We mitigate the downside of these interruptions through the use of persistence features, including optional fallback to On-Demand instances. So you can optimize workflows according to your budget and time constraints.

61589d2c5abd8e76d47f864d_download.png

Train on Cloud, Easily

Train on cloud as easily as with a local machine. We abstract away the complexity of AWS/GCP so you can iterate faster.

61589cde9047f92c62c6a419_dollar-2719007_1280.png

Cut Training Costs

We help you save up to 70% by using AWS Spot Instances or GCP Preemptible VMs.

Instance Auto-Recovery

We immediately resume training after interruptions, using the last model checkpoint via persistent EBS volume.

61589bb058645d349d682829_vertical-logo-monochromatic.png

Develop with Docker

Develop your model locally using a Docker container and run it on cloud.

Pricing

Open Source

Non-commercial use
Commercial use for a single user per account
Up to 200 hours of free spot training per month
Support through GitHub

Business

$9.99/month

Commercial use for more than 1 user per account
Email/Phone support
> 200 hours of spot training per month
Priority feature requests

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK