Document detail
Indexing facts, a per-file chunking override, the ingestion plan and the chunk list.
Click a document's name in the table to open its detail page. Three buttons at the top:
Original (download), Re-index, Stop (while running). If the document failed, a red
alert at the top states why.
Ingestion
The stat list: Chunks, Tokens, Size, Uploaded, Indexed. Below it, the chunking method in force for this file.
Chunking override
One file can use a different method from its knowledge base — say the knowledge base is on General but an FAQ file should be Q&A. Pick it in the Chunking method field (Knowledge base default removes the override), then Re-index.
Ingestion plan
Ragenta splits a long file into page ranges and processes each as a task. This table lists
every task with its status pending · running · done · failed · reused · cancelled.
reused means that page range had the same content and the same settings as last time, so the
previous result was kept — no credits spent. That is why re-indexing after only changing
keywords is far cheaper than uploading a new file.
Chunks
The passages in order, paginated. Each shows:
- its ordinal and kind:
passage(ordinary),qa(question-answer pair),row(table row),summary(produced by the Summary tree); - page range and token count;
- keywords and questions when enrichment is on.
This is the fastest way to check that a chunking method cuts in the right places before uploading a hundred files.