paper_type.json (306B)
1 { 2 "paper_type": "empirical", 3 "reason": "Conducts experiments measuring vulnerability rates in LLM-generated code (9.8%-42.1%) and quantifies the effectiveness of different guidance strategies, with the primary contribution being empirical findings on code generation flaws and mitigation techniques." 4 }