25 lines
727 B
JSON
25 lines
727 B
JSON
{
|
|
"input": "ModelTraining/ClipboardSemantics/open-training-corpus.jsonl",
|
|
"languageCounts": {
|
|
"en": 240,
|
|
"zh-Hans": 120
|
|
},
|
|
"promptSHA256": "4bd3dbc01202f2a0c69c77ee30c66904889a31485fed3bc1944b286f9ab74cff",
|
|
"promptVersion": "clipboard-consensus-v1",
|
|
"queue": "ModelTraining/ClipboardSemantics/Consensus/labeling-queue.jsonl",
|
|
"recordCount": 360,
|
|
"schemaVersion": 1,
|
|
"seed": 20260827,
|
|
"sourceCounts": {
|
|
"AmazonScience/MASSIVE en-US": 40,
|
|
"AmazonScience/MASSIVE zh-CN": 40,
|
|
"CFPB via hpe-ai/customer-complaints": 40,
|
|
"CLINC150 UCI": 40,
|
|
"CrossWOZ": 40,
|
|
"Google Research GoEmotions": 40,
|
|
"Google Taskmaster-1": 40,
|
|
"Meituan-Dianping/ASAP": 40,
|
|
"MultiDoGO": 40
|
|
}
|
|
}
|