Understanding the behavior of the Training feature
Training your AI Block is the first step to making predictions, and there's a lot that happens behind the scenes in this process. Here are some common outcomes, and what you can do about them.
My training run says it failed, what do I do now?
Training is a complex mechanism that requires the use of a few different processes. Infrequently, these processes may not perform as they should, causing incorrect behavior in the training sequence. Rather than compromise your training score, the training process will abort, and you will receive a message that it has failed. You will not be charged a training credit if your run fails.
If you receive this message, you may try again by clicking the Re-Train button above the message. Failed training runs typically succeed on the next attempt.
My training run succeeded, but now I have a bad Performance score. What can I do?
Performance is how we measure how well we think the AI Block will perform on real data. There are a few reasons why you might get a bad score after running your training, and a few ways you can improve it based on the cause. We recommend reviewing your training data to make sure you have a decent amount of balanced, good-quality training data. Before attempting to re-train the AI Block, perform your own tests on the Block from our Test tab and note the results.
If you are still unsure about the outcome of your training run, reach out to our support team for assistance.