Chunker
Represent text splitting between document loading and embedding.
Category: Processing. Registry label: Chunker.
Purpose #
Chunking splits larger text into smaller passages so retrieval can operate on manageable units. The appropriate strategy depends on document structure and the question patterns an application must support.
Position in the graph #
Document Loader → Chunker → Embedder
Design guidance #
Use this node after Document Loader and before Embedder. Open its configuration dialog to inspect the actual controls available in the current editor. When reasoning about a future runtime, consider the relationship between chunk boundaries, overlap, and retrieval quality, but do not assume a specific splitter implementation is active.
Current limitation #
No document bytes are processed into real chunks by the frontend preview.