paper_type.json (264B)
1 { 2 "paper_type": "empirical", 3 "reason": "Conducts controlled experiments testing prompt format effects across multiple models (GPT-3.5, GPT-4) on existing benchmarks (code translation, MMLU), with quantitative performance metrics as the primary contribution." 4 }