Training data from a collection: llmgw ctl kb dataset

A chat model of the cluster reads each chunk and writes questions a person
might ask of it with the answers it holds, streamed as NDJSON a chunk at a
time, for people to correct before an adapter is trained on it. A chunk that
gave no pairs says so on its own line.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Ilker ManapandClaude Opus 5.5 committed 2026-10-08 20:04:37 +02:00
1 parent 0a6a586dfe
commit ae95ac8261
8 files changed
+455

No files matched your search

+3
View File
@@ -288,6 +288,9 @@ var Routes = []Route{
Description: "On an organization, team, project or key; the narrowest one with a default wins. No collections clears it. " +
"A request's own X-Llmgw-Knowledge overrides it, and X-Llmgw-Knowledge: none turns it off.",
Request: KBDefaultRequest{}, Response: KBDefaultView{}},
{Method: "POST", Path: "/api/v1/kb/collections/{id}/dataset", Summary: "Training data from a collection: questions and answers a chat model writes from each chunk",
Description: "NDJSON, a line per pair as each chunk is read, or a line with error for a chunk that gave none. For people to correct before an adapter is trained on it.",
Request: KBDatasetRequest{}, Response: KBDatasetLine{}},
{Method: "POST", Path: "/api/v1/kb/search", Summary: "Try a search of a collection, as an administrator",
Description: "No tenancy applies: this is for checking what a collection holds and finds.",
Request: KBSearchRequest{}, Response: []KBHitView{}},