In this context, training refers to feeding ViaSay’s AI all of the training phrases found in the AI-based intents of certain User Actions. This enables the AI to digest the information and better comprehend the patterns underlying each intent.

This is important to know when trying to figure out why a training failed.

1 - How to notice that the training has failed?

The training status tab in the sidebar shows the training status of different User Actions available in your bot:

• Queued: training request is well received and training will begin soon
• Running: training has started and is in progress
• Completed: training request has been successfully completed
• Failed: training request had an error. See the column "Details" to understand why it failed.

There's also a special status called "rule-based" that doesn't require any training since Rule-based intents are triggered by simple rules and keywords rather than ViaSay AI.

2 - What are the types of errors?

The Details column tells you why a given training failed. Reasons for failure include:

• some intents in the trained User Action do not have enough training phrases
• entities linked to some intents may have been incorrectly linked or deployed

3a - How to fix the "Missing examples" error?

The most common cause of a training error in your bot is when you do not provide enough training phrases. As a reminder, you need a strict minimum of 10 training sentences in your intent for it to be trained.

3b - How to fix the "Intent is not trained" error?

1. Go to the training section

2. Click on the details of the failed training and add the missing samples.

3. Go to the intent in another tab and select the user action that needs training. You can retrain the intents of the user action by clicking on the arrow on the right hand side.

You can also retrain the whole bot by going to the main view the your use cases and clicking on the "Re-train" button