TextInsightBench / release.json
erwinmsmith's picture
Rebuild open exploration tasks and corpus-grounded evaluation
149da17 verified
Raw
History Blame Contribute Delete
1.02 kB
{
"version": "textinsightbench",
"tasks": 50,
"evaluation_documents": 435000,
"learning_documents": 944468,
"learning_by_source": {
"amazon_beauty": 330885,
"app_reviews": 1660,
"cfpb": 483788,
"nhtsa": 128135
},
"evaluation_by_source": {
"amazon_beauty": {
"tasks": 13,
"documents": 130000
},
"app_reviews": {
"tasks": 13,
"documents": 65000
},
"cfpb": {
"tasks": 12,
"documents": 120000
},
"nhtsa": {
"tasks": 12,
"documents": 120000
}
},
"task_family_counts": {
"group_difference": 20,
"temporal_change": 15,
"compound_association": 15
},
"fixed_reference_conclusions": 0,
"scoring_version": "finding-quality-discovery",
"split_policy": "Disjoint task IDs and remaining curated learning pool; inherited normalized/template deduplication.",
"historical_exposure": "Task documents were drawn from a previously public learning pool. Not an unseen or independent validation set."
}