Naming Change

deepset Cloud is now deepset AI Platform, or deepset for short. We're gradually updating the name across our product interface, documentation, and website. During this transition period, you may still see references to deepset Cloud. Bear with us while we complete this change.

Easier Component Connections

We've improved suggestions for component connections in Pipeline Builder. Now, when you click a connection point, you can see a list of popular and compatible components. Hover over a suggestion to view a tooltip with details about what the component does. All this to help you choose the most efficient connections for your pipelines.

For details, see Create a Pipeline in Pipeline Builder.

Runtime Parameters in Jobs and Playground

The query set for jobs now has the params column where you can provide runtime parameters for components in your pipeline. These parameters override the parameters set in the pipeline and are only used for queries from the query set. For details, see Prepare a Query Set.

You can also change the parameters for individual components in Playground. That's the fastest and easiest way to test your pipeline configuration. Click the Configurations button in Playground, update components' parameters and ask a query.

For details, see Modify Pipeline Parameters at Query Time.

Streaming

We added the streaming_callback parameter to the Generators, allowing you to configure the LLMs in your pipeline to stream. To enable streaming, set their streaming_callback to deepset_cloud_custom_nodes.callbacks.streaming.streaming_callback. If your pipeline has multiple Generators, you can enable streaming for one or all of them.

For details, see Generators.