LLM-generated C translations of small Python programs can be verified by CBMC in 80 to 92 percent of benchmark cases, but fault localization stays below 53 percent.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
PyVeritas: On Verifying Python via LLM-Based Transpilation and Bounded Model Checking for C
LLM-generated C translations of small Python programs can be verified by CBMC in 80 to 92 percent of benchmark cases, but fault localization stays below 53 percent.