Thank you for your work on Docling!
The /v1/chunk/hybrid/file/async (and /v1/chunk/hybrid/file) endpoints expose several HybridChunker options but repeat_table_header is not available.
When using HybridChunker directly, this can be set to False to avoid inflating chunk count for documents with large tables. In particular in our app we contextualize seperately so the table headers are undesirable.
Env:
- docling-serve: 1.15.0
- docling-core: 2.70.2
Thank you for your work on Docling!
The
/v1/chunk/hybrid/file/async(and/v1/chunk/hybrid/file) endpoints expose several HybridChunker options butrepeat_table_headeris not available.When using
HybridChunkerdirectly, this can be set toFalseto avoid inflating chunk count for documents with large tables. In particular in our app we contextualize seperately so the table headers are undesirable.Env: