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>

docs/admin-api.openapi.json

@@ -1185,6 +1185,58 @@
         }
       }
     },
+    "/api/v1/kb/collections/{id}/dataset": {
+      "post": {
+        "operationId": "postKbCollectionsIdDataset",
+        "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.",
+        "tags": [
+          "kb"
+        ],
+        "parameters": [
+          {
+            "name": "id",
+            "in": "path",
+            "required": true,
+            "description": "The collection's ID, its name when that is unique, or ORG/NAME.",
+            "schema": {
+              "type": "string"
+            }
+          },
+          {
+            "$ref": "#/components/parameters/Actor"
+          }
+        ],
+        "requestBody": {
+          "required": true,
+          "content": {
+            "application/json": {
+              "schema": {
+                "$ref": "#/components/schemas/KBDatasetRequest"
+              }
+            }
+          }
+        },
+        "responses": {
+          "200": {
+            "description": "OK",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/KBDatasetLine"
+                }
+              }
+            }
+          },
+          "401": {
+            "$ref": "#/components/responses/Unauthorized"
+          },
+          "default": {
+            "$ref": "#/components/responses/Error"
+          }
+        }
+      }
+    },
     "/api/v1/kb/collections/{id}/documents": {
       "get": {
         "operationId": "getKbCollectionsIdDocuments",
@@ -5059,6 +5111,45 @@
           }
         }
       },
+      "KBDatasetLine": {
+        "type": "object",
+        "description": "KBDatasetLine is one line of the answer: a pair, or why a chunk gave none.",
+        "properties": {
+          "q": {
+            "type": "string"
+          },
+          "a": {
+            "type": "string"
+          },
+          "doc": {
+            "type": "string"
+          },
+          "ord": {
+            "type": "integer"
+          },
+          "error": {
+            "type": "string"
+          }
+        }
+      },
+      "KBDatasetRequest": {
+        "type": "object",
+        "description": "KBDatasetRequest asks for training data from a collection.",
+        "properties": {
+          "model": {
+            "type": "string",
+            "description": "Model writes the questions and answers: a chat model of the catalog."
+          },
+          "per_chunk": {
+            "type": "integer",
+            "description": "PerChunk is how many pairs each chunk gives; 8 when 0."
+          },
+          "max_chunks": {
+            "type": "integer",
+            "description": "MaxChunks bounds how many chunks are read; 500 when 0."
+          }
+        }
+      },
       "KBDefaultRequest": {
         "type": "object",
         "description": "KBDefaultRequest sets the collections a part of the tree answers from when a request names none; none clears it.",