PH — The two benchmark tasks and datasets were locked
The two benchmark tasks and datasets were locked: issue triage on the Jira Social Repository (Apache, multiclass priority) and story-point estimation on TAWOS, with success thresholds F1-macro >= 0.55 and MAE ⇐ 3.0 SP.
Source: PUMA project documentation · Traceability: corpus unit
PHASE-010· Confidence: verified-at-primary-source