CodeSpecBench shows LLMs achieve at most 20.2% pass rate on repository-level executable behavioral specification generation, revealing that strong code generation does not imply deep semantic understanding.
Evocodebench: An evolving code generation benchmark with domain-specific evaluations
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
CodeSpecBench: Benchmarking LLMs for Executable Behavioral Specification Generation
CodeSpecBench shows LLMs achieve at most 20.2% pass rate on repository-level executable behavioral specification generation, revealing that strong code generation does not imply deep semantic understanding.