Tracking your experiments has never been easier with the new cnvrg.io Slack integration. cnvrg.io will now send direct messages to Slack when your experiment is successful, notify you when it's had an error, is idle, or is restarting.
Building and training machine learning models can require hundreds of experiments. You need to try new ideas, algorithms and parameters to get your model to converge. cnvrg.io has a great feature that lets you run 100’s of experiments in a single command with Grid-Search. This is a great tool for hyperparameter optimization. You may also have some long-time running experiments that may run for a few weeks.
Now you can get live updates on the progress of your experiments. Every time your experiment changes status you’ll automatically receive a friendly message through Slack. You’ll get notified when:
The cnvrg.io Slack integration will also notify you as you’re saving in cloud costs! With cnvrg.io spot instances integration you can save up to 80% in cloud costs. Spot instances allow users to utilize unused machines temporarily at a majorly reduced price. Once the Spot is needed, AWS will terminate the instance. Luckily, cnvrg.io will automatically relaunch training on a new instance, using the latest checkpoint without you lifting a finger. With that said, it’s nice to know when business is happening to your model. So, we introduced Slack notifications for the following Spot instance changes:
Now, you can experience the emotional rollercoaster along with your precious experiments! Consider these machine learning experiment Slack notifications your baby monitor. You’ll know what's going on with your experiments at every step of the way. Don’t worry - all your experiments are in good hands.
Bonus tip: Did you know that you can change the sound of your Slack notifications to ‘Hummus’? We recommend this as your default for all cnvrg.io notifications.