FAQs for watsonx Code Assistant

FAQs for watsonx Code Assistant might include questions about code implementation, quality, or platforms.

How is the `pom.xml` file of a Maven project located?

watsonx Code Assistant Standard plan

Watsonx Code Assistant uses the aggregator pom.xml file to build and manage the entire multi-module Maven project. When watsonx Code Assistant attempts to do builds and other Maven-related activity, it uses the multi-module root (MMR) to locate the aggregator pom.xml file.

The MMR first searches the highest level project directory for the aggregator pom.xml file.
If the MMR doesn't find the file, it searches the project directory by going from the highest to the next highest directory structure, and so on, until it finds an aggregator pom.xml file.
If the MMR again does not find an aggregator pom.xml file, it searches the project directory by going from the highest to the next highest directory structure, and so on, until it finds a regular pom.xml file. A regular pom.xml file indicates that the Maven project is a single module project instead of a multi-module project.

Why is a .class file included in an analysis report instead of the .java file?

watsonx Code Assistant Standard plan

When you run an analysis of a binary (JAR/WAR/EAR), the issues are always discovered in the class files. When the binary scanner is run outside of watsonx Code Assistant, it has no access to the source code, so the .class file is included in the analysis report. When you import a migration plan, or when you run an analysis inside watsonx Code Assistant, the reports show references to class files, but watsonx Code Assistant links the .class file to the .java file, and the .java files are shown in the watsonx Code Assistant UI.

What data sources are used to train the watsonx Code Assistant models?

The watsonx Code Assistant models gather training data from various sources depending on which platform it's supporting. For more information, see:

Can I train or tune the model?

Red Hat Ansible Lightspeed

If you purchased an IBM watsonx Code Assistant for Red Hat Ansible Lightspeed Standard plan, you can tune the IBM base code model on your data so it generates code suggestions that are customized for your enterprise standards. You can use the watsonx Code Assistant for Red Hat Ansible Lightspeed tuning studio to create model experiments and deploy your models to shared spaces so you and your team can quickly generate reliable and accurate code. For more information, see Tuning the IBM base code model for watsonx Code Assistant for Red Hat Ansible Lightspeed.

Can I provide feedback about the model?

You can provide feedback on your experiences, including suggestions for when your results don't match your expectations. For more information about providing feedback, see the IBM Data and AI Ideas Portal for Customers.

How long does it take before I see my feedback improve the model?

Training the IBM watsonx Code Assistant model is resource-intensive. IBM Research intends to retrain the model at a cadence that provides noticeable model improvements between model versions.