Our team has used the last week of February to release a number of improvements on the product. Take a look at what's new in the past few weeks.
We have created a MindsDB Docker container that can be used on Amazon SageMaker.
Now, you can use MindsDB on SageMaker to train and deploy models within SageMaker, start endpoints and take advantage of automated explainable machine learning with MindsDB. The following code using SageMaker SDK will start the train job, deploy the model and return predictions.
For additional details please check:
The MindsDB server previously used Flask-RestPlus for building its REST API. However, responses with it were less than optimal and required frequent bug fixes sowe’ve migrated to Flask-RESTX, which is the maintained fork of Flask-RestPlus.
Our team constantly works on improving the MindsDB Scout UI/UX design to provide the best experience to our users. The latest version of MindsDB Scout, which we’ve been working heavily on, will be announced in the next few weeks.
The latest release 0.17.1 of Lightwood includes the following:
We have also added new monitoring functionality using the Visdom tool for creating and sharing visualizations of data. To install and start the Visdom server run visdom.sh located in the root directory of Lightwood.
The script will install dependencies, start the Visdom server and open the browser where the plots shall be displayed. If the server is not automatically started run:
python3 -m visdom.server
To have Lightwood report loss changes and network heatmaps go to config.py and change the values you want to monitor.
In February, we started hosting weekly Live Machine Learning classes that aim to present machine learning as a valuable skill for developers to possess. Our main goal is to help these developers learn about machine learning fundamentals so they can use ML in their projects, and also present them with the ways they can use MindsDB’s free open source framework to train models and make predictions.