paper_type.json (270B)
1 { 2 "paper_type": "benchmark-creation", 3 "reason": "GAIA introduces a new 466-question benchmark for evaluating general AI assistants; while the paper includes baseline experiments, the primary contribution is the benchmark itself and its difficulty level framework." 4 }