When you’re trying to train the best model, especially on a deadline, there should be a method to the “madness.” That’s why we built Experiment Management into SigOpt. Experiment Management, released back in July, facilitates runs and experiment tracking, so that you never lose track of the progress you made, and you can effectively decide what part of your model or training process to adjust next.
In conjunction with MLconf, Head of Product Fay Kallel, and Head of Engineering Jim Blomo will present a couple product walkthroughs, live with notebooks and API access tokens, so that you can follow along. Here’s the workflow he’ll be exploring, using a fraud detection (xgboost) and an image classification (Keras and CIFAR) example:
Jim will also take questions from the audience, mapping users’ and data scientists’ modeling workflow to our platform, and identifying which integrations with external tools would be most useful. He’ll also field questions about SigOpt’s future roadmap, and where he sees the next obvious steps in building out tools to support productive modeling workflows.
Click here to sign up for the webinar as part of MLconf’s 2020 virtual online event next week, on Tuesday, September 22, 2020 at 12:00 PM PST.
If you’re interested in trying out Experiment Management, please sign up for the beta here. If you’re interested to learn more about the product, check out the launch blog post here, or read our documentation.